Model profile · OpenAI

GPT-5.4 mini

Closed weightspremium · registry tag 2026 mini

Verified but aging

Reads as verified but aging across the resolved source data. Capability fit excludes speed and cost.

Visible coverage: 65.8%
Verified coverage: 56.8%
Spread: 98.5%
Last verified: Oct 3, 2026

44%capability fit

textcodevisiondocumentsearch18 aliases46 official source links

Open compare

Data version

Current snapshot.

Published Oct 3, 2026Model list checked4 stale sources105 claim warnings9 providers · 1334 tracked models

The registry snapshot and page stamp are shown so a stale deploy is visible at a glance. View live data status.

Source-linked scores by benchmark

Each row keeps the benchmark source, source type, raw metric, and percentile inside its fair comparison set. Capability summaries exclude archived categories, speed, and cost; those rows remain below.

Verified but agingThis model currently reads as verified but aging across the resolved source data.

Chat / text36 benchmarks43.4%

Intelligence Index

AA · Chat / text · Combined

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #298 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 10
Percentile: 46.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `intelligenceIndex`.

46.5% percentile inside its fair comparison set

10Raw benchmark value

AA-Omniscience accuracy

AA · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #235 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 19.2%
Percentile: 48.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceAccuracy`.

48.3% percentile inside its fair comparison set

19.2%Raw benchmark value

AA-Omniscience non-hallucination

AA · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #430 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 4.4%
Percentile: 5.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceNonHallucination`.

5.3% percentile inside its fair comparison set

4.4%Raw benchmark value

IFBench

AA · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #205 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 38.8%
Percentile: 41.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `ifbench`.

41.4% percentile inside its fair comparison set

38.8%Raw benchmark value

Input price

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #209 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $0.8 /1M input tokens
Percentile: 41.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mInputTokens`.

41.2% percentile inside its fair comparison set

$0.8 /1M input tokensRaw benchmark value

Output price

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #243 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $4.5 /1M output tokens
Percentile: 31.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mOutputTokens`.

31.6% percentile inside its fair comparison set

$4.5 /1M output tokensRaw benchmark value

Output Speed

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #75 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 127.1 tokens/s
Percentile: 71.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianOutputTokensPerSecond`.

71.8% percentile inside its fair comparison set

127.1 tokens/sRaw benchmark value

Time to first token

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #253 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 123.46s
Percentile: 4.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstTokenSeconds`.

4.5% percentile inside its fair comparison set

123.46sRaw benchmark value

Time to first answer token

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #251 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 123.46s
Percentile: 4.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstAnswerTokenSeconds`.

4.6% percentile inside its fair comparison set

123.46sRaw benchmark value

Text Arena

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #76 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,447
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #84. Votes: 63290. Organization: openai. License: Proprietary.

80.1% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,443 - 1,451

Text Arena · Creative Writing

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #97 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,401
Percentile: 74.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing. Source rank: #107. Votes: 10797. Organization: openai. License: Proprietary.

74.3% percentile inside its fair comparison set

1,401Raw benchmark valueCI 1,394 - 1,409

Text Arena · English

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,450
Percentile: 77.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #92. Votes: 29062. Organization: openai. License: Proprietary.

77.9% percentile inside its fair comparison set

1,450Raw benchmark valueCI 1,445 - 1,455

Text Arena · Exclude Ties

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #76 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,449
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: exclude_ties. Source rank: #84. Votes: 48664. Organization: openai. License: Proprietary.

80.1% percentile inside its fair comparison set

1,449Raw benchmark valueCI 1,444 - 1,453

Text Arena · Hard Prompts

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #75 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,469
Percentile: 80.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts. Source rank: #83. Votes: 41568. Organization: openai. License: Proprietary.

80.3% percentile inside its fair comparison set

1,469Raw benchmark valueCI 1,464 - 1,473

Text Arena · Hard Prompts English

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #87 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,466
Percentile: 77%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts_english. Source rank: #95. Votes: 19794. Organization: openai. License: Proprietary.

77% percentile inside its fair comparison set

1,466Raw benchmark valueCI 1,460 - 1,472

Text Arena · Instruction Following

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #82 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,434
Percentile: 78.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: instruction_following. Source rank: #89. Votes: 21670. Organization: openai. License: Proprietary.

78.5% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,428 - 1,439

Text Arena · Longer Query

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #88 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,447
Percentile: 75.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: longer_query. Source rank: #96. Votes: 27722. Organization: openai. License: Proprietary.

75.5% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,441 - 1,452

Text Arena · Multi Turn

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #61 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,466
Percentile: 84%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: multi_turn. Source rank: #68. Votes: 11873. Organization: openai. License: Proprietary.

84% percentile inside its fair comparison set

1,466Raw benchmark valueCI 1,459 - 1,472

Text Arena · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #123 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,411
Percentile: 67.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #135. Votes: 63290. Organization: openai. License: Proprietary.

67.6% percentile inside its fair comparison set

1,411Raw benchmark valueCI 1,408 - 1,415

Text Arena · Creative Writing · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #126 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,370
Percentile: 66.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing. Source rank: #141. Votes: 10797. Organization: openai. License: Proprietary.

66.6% percentile inside its fair comparison set

1,370Raw benchmark valueCI 1,363 - 1,377

Text Arena · English · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #136 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,412
Percentile: 64.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #150. Votes: 29062. Organization: openai. License: Proprietary.

64.1% percentile inside its fair comparison set

1,412Raw benchmark valueCI 1,407 - 1,417

Text Arena · Exclude Ties · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #123 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,397
Percentile: 67.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: exclude_ties. Source rank: #134. Votes: 48664. Organization: openai. License: Proprietary.

67.6% percentile inside its fair comparison set

1,397Raw benchmark valueCI 1,393 - 1,402

Text Arena · Hard Prompts · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #112 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,424
Percentile: 70.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts. Source rank: #123. Votes: 41568. Organization: openai. License: Proprietary.

70.5% percentile inside its fair comparison set

1,424Raw benchmark valueCI 1,420 - 1,429

Text Arena · Hard Prompts English · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #127 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,420
Percentile: 66.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts_english. Source rank: #139. Votes: 19794. Organization: openai. License: Proprietary.

66.3% percentile inside its fair comparison set

1,420Raw benchmark valueCI 1,415 - 1,426

Text Arena · Instruction Following · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #105 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,405
Percentile: 72.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: instruction_following. Source rank: #114. Votes: 21670. Organization: openai. License: Proprietary.

72.3% percentile inside its fair comparison set

1,405Raw benchmark valueCI 1,399 - 1,410

Text Arena · Longer Query · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #121 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,407
Percentile: 66.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: longer_query. Source rank: #133. Votes: 27722. Organization: openai. License: Proprietary.

66.2% percentile inside its fair comparison set

1,407Raw benchmark valueCI 1,402 - 1,413

Text Arena · Multi Turn · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #92 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,429
Percentile: 75.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: multi_turn. Source rank: #102. Votes: 11873. Organization: openai. License: Proprietary.

75.7% percentile inside its fair comparison set

1,429Raw benchmark valueCI 1,422 - 1,436

Instruction following

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #59 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 59.8%
Percentile: 10.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: IF. Tasks scored: 4.

10.8% percentile inside its fair comparison set

59.8%Raw benchmark value

Language

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #61 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 71%
Percentile: 7.7%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Language. Tasks scored: 3.

7.7% percentile inside its fair comparison set

71%Raw benchmark value

Paraphrase

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #55 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 58.9%
Percentile: 16.9%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: paraphrase. Category: IF.

16.9% percentile inside its fair comparison set

58.9%Raw benchmark value

Simplify

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #60 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 54.1%
Percentile: 9.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: simplify. Category: IF.

9.2% percentile inside its fair comparison set

54.1%Raw benchmark value

Story generation

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #55 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 66.6%
Percentile: 16.9%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: story_generation. Category: IF.

16.9% percentile inside its fair comparison set

66.6%Raw benchmark value

Summarize

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #58 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 59.6%
Percentile: 12.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: summarize. Category: IF.

12.3% percentile inside its fair comparison set

59.6%Raw benchmark value

Connections

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #33 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 99%
Percentile: 50.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: connections. Category: Language.

50.8% percentile inside its fair comparison set

99%Raw benchmark value

Plot unscrambling

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 43.9%
Percentile: 3.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: plot_unscrambling. Category: Language.

3.1% percentile inside its fair comparison set

43.9%Raw benchmark value

Typos

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #60 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 70%
Percentile: 10.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: typos. Category: Language.

10.8% percentile inside its fair comparison set

70%Raw benchmark value

Coding30 benchmarks44.8%

Terminal-Bench Hard

AA · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #135 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 14.4%
Percentile: 59.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `terminalbenchHard`.

59.6% percentile inside its fair comparison set

14.4%Raw benchmark value

SciCode

AA · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #147 · Source label: GPT-5 mini (High)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 39%
Percentile: 26.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `scicode`.

26.3% percentile inside its fair comparison set

39%Raw benchmark value

Text Arena · Coding

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #79 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,495
Percentile: 79%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: coding. Source rank: #85. Votes: 17210. Organization: openai. License: Proprietary.

79% percentile inside its fair comparison set

1,495Raw benchmark valueCI 1,489 - 1,501

Text Arena · Coding · No Style Control

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #110 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,438
Percentile: 70.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: coding. Source rank: #119. Votes: 17210. Organization: openai. License: Proprietary.

70.6% percentile inside its fair comparison set

1,438Raw benchmark valueCI 1,432 - 1,444

Vibe Code Bench v1.1

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #46 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 48%
Percentile: 51.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: vibe-code; provider: OpenAI.

51.6% percentile inside its fair comparison set

48%Raw benchmark valueCI 37% - 59%

WebDev Arena

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,397
Percentile: 34.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #87. Votes: 11998. Organization: openai. License: Proprietary.

34.6% percentile inside its fair comparison set

1,397Raw benchmark valueCI 1,391 - 1,404

Code Arena · Webdev Html

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #76 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,420
Percentile: 40%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: webdev-html. Source rank: #81. Votes: 1437. Organization: openai. License: Proprietary.

40% percentile inside its fair comparison set

1,420Raw benchmark valueCI 1,404 - 1,437

Code Arena · Webdev React

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #81 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,386
Percentile: 29.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: webdev-react. Source rank: #86. Votes: 9878. Organization: openai. License: Proprietary.

29.2% percentile inside its fair comparison set

1,386Raw benchmark valueCI 1,379 - 1,393

Code Migration

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #54 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 12.9%
Percentile: 22.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: code-migration; provider: OpenAI.

22.1% percentile inside its fair comparison set

12.9%Raw benchmark valueCI 5.8% - 20.1%

ProgramBench

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #39 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 0%
Percentile: 51.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: programbench; provider: OpenAI.

51.7% percentile inside its fair comparison set

0%Raw benchmark valueCI 0% - 0%

Terminal-Bench 2.1

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #45 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 54.7%
Percentile: 34.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: terminal-bench-2-1; provider: OpenAI.

34.3% percentile inside its fair comparison set

54.7%Raw benchmark valueCI 45.8% - 63.6%

SWE-bench Verified

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #49 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 73%
Percentile: 41.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: swebench; provider: OpenAI.

41.5% percentile inside its fair comparison set

73%Raw benchmark valueCI 69.1% - 76.9%

LiveCodeBench

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #59 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 81.5%
Percentile: 53.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: lcb; provider: OpenAI.

53.6% percentile inside its fair comparison set

81.5%Raw benchmark valueCI 79.3% - 83.6%

Agentic coding

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #58 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 41.7%
Percentile: 12.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Agentic Coding. Tasks scored: 3.

12.3% percentile inside its fair comparison set

41.7%Raw benchmark value

Coding

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #56 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 71.6%
Percentile: 15.4%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Coding. Tasks scored: 2.

15.4% percentile inside its fair comparison set

71.6%Raw benchmark value

JavaScript

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 50%
Percentile: 3.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: javascript. Category: Agentic Coding.

3.1% percentile inside its fair comparison set

50%Raw benchmark value

TypeScript

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #63 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 20%
Percentile: 7.7%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: typescript. Category: Agentic Coding.

7.7% percentile inside its fair comparison set

20%Raw benchmark value

Python

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #27 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 55%
Percentile: 63.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: python. Category: Agentic Coding.

63.1% percentile inside its fair comparison set

55%Raw benchmark value

Coding generation

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 67.7%
Percentile: 3.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: code_generation. Category: Coding.

3.1% percentile inside its fair comparison set

67.7%Raw benchmark value

Coding completion

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #49 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 75.6%
Percentile: 26.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: code_completion. Category: Coding.

26.2% percentile inside its fair comparison set

75.6%Raw benchmark value

Debugging

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #14 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 84.1%
Percentile: 50%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-debugging.

50% percentile inside its fair comparison set

84.1%Raw benchmark value

Refactoring

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #13 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 62.3%
Percentile: 42.9%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-refactoring.

42.9% percentile inside its fair comparison set

62.3%Raw benchmark value

Security

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #5 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 83.3%
Percentile: 84%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-security.

84% percentile inside its fair comparison set

83.3%Raw benchmark value

UI bench

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #9 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 72.6%
Percentile: 63.6%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ui.

63.6% percentile inside its fair comparison set

72.6%Raw benchmark value

Speed throughput

BB · Coding · Speed / cost

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #4 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 236.4 t/s
Percentile: 88%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-throughput.

88% percentile inside its fair comparison set

236.4 t/sRaw benchmark value

Speed TTFT

BB · Coding · Speed / cost

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #1 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 233.00ms
Percentile: 100%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ttft.

100% percentile inside its fair comparison set

233.00msRaw benchmark value

Speed cost

BB · Coding · Speed / cost

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #7 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: $0.1 $
Percentile: 76%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-speed-cost.

76% percentile inside its fair comparison set

$0.1 $Raw benchmark value

Overall quality

BB · Coding · Combined

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #11 · Source label: GPT-5.4 Mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 77.6%
Percentile: 54.5%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-overall-quality.

54.5% percentile inside its fair comparison set

77.6%Raw benchmark value

Overall vibe

BB · Coding · Combined

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #11 · Source label: GPT-5.4 Mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 77.5%
Percentile: 54.5%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-overall-vibe.

54.5% percentile inside its fair comparison set

77.5%Raw benchmark value

Overall coverage

BB · Coding · Combined

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #9 · Source label: GPT-5.4 Mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 100%
Percentile: 100%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-overall-coverage.

100% percentile inside its fair comparison set

100%Raw benchmark value

Reasoning / math / science20 benchmarks32.5%

Humanity's Last Exam

AA · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #356 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 5.1%
Percentile: 32.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `hle`.

32.9% percentile inside its fair comparison set

5.1%Raw benchmark value

GPQA

AA · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #293 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 60.6%
Percentile: 40.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gpqa`.

40.8% percentile inside its fair comparison set

60.6%Raw benchmark value

CritPt

AA · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #229 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 0%
Percentile: 50.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `critpt`.

50.1% percentile inside its fair comparison set

0%Raw benchmark value

TutorBench

SL · Reasoning / math / science · Rubric

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #5 · Source label: gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 55.3%
Percentile: 88.9%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from the public Scale Labs page for scale-tutorbench. Reported model configuration: gpt-5-2025-08-07. Collapse policy: highest reported score per canonical model. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

88.9% percentile inside its fair comparison set

55.3%Raw benchmark value

MultiNRC

SL · Reasoning / math / science · Rubric

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #3 · Source label: gpt-5-pro-2025-10-06

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 65.2%
Percentile: 97.7%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from the public Scale Labs page for scale-multinrc. Reported model configuration: gpt-5-pro-2025-10-06. Collapse policy: highest reported score per canonical model. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

97.7% percentile inside its fair comparison set

65.2%Raw benchmark value

Text Arena · Math

AR · Reasoning / math / science · Human

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,439
Percentile: 76.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: math. Source rank: #91. Votes: 3361. Organization: openai. License: Proprietary.

76.9% percentile inside its fair comparison set

1,439Raw benchmark valueCI 1,428 - 1,450

Text Arena · Math · No Style Control

AR · Reasoning / math / science · Human

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #115 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,418
Percentile: 68.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: math. Source rank: #124. Votes: 3361. Organization: openai. License: Proprietary.

68.2% percentile inside its fair comparison set

1,418Raw benchmark valueCI 1,407 - 1,429

GPQA Diamond

VALS-AI · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #59 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 83.1%
Percentile: 53.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

53.2% percentile inside its fair comparison set

83.1%Raw benchmark valueCI 78.3% - 87.9%

MMLU Pro

VALS-AI · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #63 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 84.6%
Percentile: 50.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: mmlu_pro; provider: OpenAI.

50.4% percentile inside its fair comparison set

84.6%Raw benchmark valueCI 83.9% - 85.3%

Mathematics

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 78.5%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Mathematics. Tasks scored: 4.

6.2% percentile inside its fair comparison set

78.5%Raw benchmark value

Reasoning

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 71.3%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Reasoning. Tasks scored: 4.

6.2% percentile inside its fair comparison set

71.3%Raw benchmark value

AMPS Hard

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #47 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 97%
Percentile: 32.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: AMPS_Hard. Category: Mathematics.

32.3% percentile inside its fair comparison set

97%Raw benchmark value

Integrals with game

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 37%
Percentile: 4.6%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: integrals_with_game. Category: Mathematics.

4.6% percentile inside its fair comparison set

37%Raw benchmark value

Math competition

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #50 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 94.1%
Percentile: 26.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: math_comp. Category: Mathematics.

26.2% percentile inside its fair comparison set

94.1%Raw benchmark value

Olympiad

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #58 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 85.7%
Percentile: 12.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: olympiad. Category: Mathematics.

12.3% percentile inside its fair comparison set

85.7%Raw benchmark value

Theory of mind

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #44 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 76.9%
Percentile: 33.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: theory_of_mind. Category: Reasoning.

33.8% percentile inside its fair comparison set

76.9%Raw benchmark value

Zebra puzzle

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 60.4%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: zebra_puzzle. Category: Reasoning.

6.2% percentile inside its fair comparison set

60.4%Raw benchmark value

Spatial

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #65 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 88%
Percentile: 1.5%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: spatial. Category: Reasoning.

1.5% percentile inside its fair comparison set

88%Raw benchmark value

Logic with navigation

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #65 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 60%
Percentile: 1.5%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: logic_with_navigation. Category: Reasoning.

1.5% percentile inside its fair comparison set

60%Raw benchmark value

Reasoning

BB · Reasoning / math / science · Rubric

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #17 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 36.2%
Percentile: 23.8%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-reasoning.

23.8% percentile inside its fair comparison set

36.2%Raw benchmark value

Professional reasoning37 benchmarks40.6%

APEX-Agents-AA

AA · Professional reasoning · Objective

Long-horizon agentic task completion.

Rank #14 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 28.2%
Percentile: 59.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `apexAgents`.

59.4% percentile inside its fair comparison set

28.2%Raw benchmark value

PRBench Legal

SL · Professional reasoning · Rubric

Applied legal reasoning on professional-domain tasks.

Rank #9 · Source label: gpt-5-pro

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 49.9%
Percentile: 82.9%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from the public Scale Labs page for scale-prbench-legal. Reported model configuration: gpt-5-pro. Collapse policy: highest reported score per canonical model. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

82.9% percentile inside its fair comparison set

49.9%Raw benchmark value

Text Arena · Expert

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena expert leaderboard.

Rank #69 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,480
Percentile: 79.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: expert. Source rank: #72. Votes: 6299. Organization: openai. License: Proprietary.

79.2% percentile inside its fair comparison set

1,480Raw benchmark valueCI 1,472 - 1,489

Text Arena · Industry Business And Management And Financial Operations

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_business_and_management_and_financial_operations leaderboard.

Rank #57 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,460
Percentile: 84.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_business_and_management_and_financial_operations. Source rank: #61. Votes: 12827. Organization: openai. License: Proprietary.

84.8% percentile inside its fair comparison set

1,460Raw benchmark valueCI 1,453 - 1,466

Text Arena · Industry Entertainment And Sports And Media

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_entertainment_and_sports_and_media leaderboard.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,409
Percentile: 77.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_entertainment_and_sports_and_media. Source rank: #92. Votes: 13880. Organization: openai. License: Proprietary.

77.8% percentile inside its fair comparison set

1,409Raw benchmark valueCI 1,402 - 1,415

Text Arena · Industry Legal And Government

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_legal_and_government leaderboard.

Rank #72 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,457
Percentile: 79.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_legal_and_government. Source rank: #82. Votes: 5040. Organization: openai. License: Proprietary.

79.7% percentile inside its fair comparison set

1,457Raw benchmark valueCI 1,448 - 1,466

Text Arena · Industry Life And Physical And Social Science

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_life_and_physical_and_social_science leaderboard.

Rank #77 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,465
Percentile: 79.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_life_and_physical_and_social_science. Source rank: #84. Votes: 10493. Organization: openai. License: Proprietary.

79.7% percentile inside its fair comparison set

1,465Raw benchmark valueCI 1,458 - 1,472

Text Arena · Industry Mathematical

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_mathematical leaderboard.

Rank #75 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,452
Percentile: 79.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_mathematical. Source rank: #81. Votes: 3420. Organization: openai. License: Proprietary.

79.2% percentile inside its fair comparison set

1,452Raw benchmark valueCI 1,441 - 1,463

Text Arena · Industry Medicine And Healthcare

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_medicine_and_healthcare leaderboard.

Rank #95 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,451
Percentile: 72.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_medicine_and_healthcare. Source rank: #108. Votes: 4757. Organization: openai. License: Proprietary.

72.8% percentile inside its fair comparison set

1,451Raw benchmark valueCI 1,441 - 1,460

Text Arena · Industry Software And It Services

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_software_and_it_services leaderboard.

Rank #77 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,485
Percentile: 79.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_software_and_it_services. Source rank: #84. Votes: 24993. Organization: openai. License: Proprietary.

79.8% percentile inside its fair comparison set

1,485Raw benchmark valueCI 1,480 - 1,490

Text Arena · Industry Writing And Literature And Language

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_writing_and_literature_and_language leaderboard.

Rank #79 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,423
Percentile: 79.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_writing_and_literature_and_language. Source rank: #88. Votes: 15820. Organization: openai. License: Proprietary.

79.2% percentile inside its fair comparison set

1,423Raw benchmark valueCI 1,417 - 1,429

Text Arena · Expert · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena expert leaderboard.

Rank #99 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,435
Percentile: 70%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: expert. Source rank: #105. Votes: 6299. Organization: openai. License: Proprietary.

70% percentile inside its fair comparison set

1,435Raw benchmark valueCI 1,426 - 1,443

Text Arena · Industry Business And Management And Financial Operations · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_business_and_management_and_financial_operations leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,418
Percentile: 74.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_business_and_management_and_financial_operations. Source rank: #103. Votes: 12827. Organization: openai. License: Proprietary.

74.3% percentile inside its fair comparison set

1,418Raw benchmark valueCI 1,411 - 1,424

Text Arena · Industry Entertainment And Sports And Media · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_entertainment_and_sports_and_media leaderboard.

Rank #124 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,373
Percentile: 67.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_entertainment_and_sports_and_media. Source rank: #136. Votes: 13880. Organization: openai. License: Proprietary.

67.1% percentile inside its fair comparison set

1,373Raw benchmark valueCI 1,366 - 1,379

Text Arena · Industry Legal And Government · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_legal_and_government leaderboard.

Rank #128 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,415
Percentile: 63.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_legal_and_government. Source rank: #141. Votes: 5040. Organization: openai. License: Proprietary.

63.6% percentile inside its fair comparison set

1,415Raw benchmark valueCI 1,406 - 1,424

Text Arena · Industry Life And Physical And Social Science · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_life_and_physical_and_social_science leaderboard.

Rank #131 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,417
Percentile: 65.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_life_and_physical_and_social_science. Source rank: #142. Votes: 10493. Organization: openai. License: Proprietary.

65.2% percentile inside its fair comparison set

1,417Raw benchmark valueCI 1,410 - 1,424

Text Arena · Industry Mathematical · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_mathematical leaderboard.

Rank #110 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,428
Percentile: 69.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_mathematical. Source rank: #117. Votes: 3420. Organization: openai. License: Proprietary.

69.4% percentile inside its fair comparison set

1,428Raw benchmark valueCI 1,417 - 1,439

Text Arena · Industry Medicine And Healthcare · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_medicine_and_healthcare leaderboard.

Rank #143 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,395
Percentile: 59%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_medicine_and_healthcare. Source rank: #159. Votes: 4757. Organization: openai. License: Proprietary.

59% percentile inside its fair comparison set

1,395Raw benchmark valueCI 1,385 - 1,404

Text Arena · Industry Software And It Services · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_software_and_it_services leaderboard.

Rank #115 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,436
Percentile: 69.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_software_and_it_services. Source rank: #125. Votes: 24993. Organization: openai. License: Proprietary.

69.7% percentile inside its fair comparison set

1,436Raw benchmark valueCI 1,430 - 1,441

Text Arena · Industry Writing And Literature And Language · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_writing_and_literature_and_language leaderboard.

Rank #113 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,392
Percentile: 70.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_writing_and_literature_and_language. Source rank: #123. Votes: 15820. Organization: openai. License: Proprietary.

70.1% percentile inside its fair comparison set

1,392Raw benchmark valueCI 1,386 - 1,398

Vals Index

VALS-AI · Professional reasoning · Combined

Weighted model performance across economically relevant Vals tasks.

Rank #38 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 33
Percentile: 7.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: vals_index; provider: OpenAI.

7.5% percentile inside its fair comparison set

33Raw benchmark valueCI 31 - 36

Legal Research Bench

VALS-AI · Professional reasoning · Objective

Applied legal research tasks.

Rank #59 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 12.5%
Percentile: 14.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: legal_research; provider: OpenAI.

14.7% percentile inside its fair comparison set

12.5%Raw benchmark valueCI 8% - 17%

Harvey's Legal Agent Benchmark

VALS-AI · Professional reasoning · Objective

Completing legal work with documents, spreadsheets, presentations, and file-system tools.

Rank #67 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 0%
Percentile: 20.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: hlab; provider: OpenAI.

20.3% percentile inside its fair comparison set

0%Raw benchmark valueCI 0% - 0%

LegalBench

VALS-AI · Professional reasoning · Objective

Academic legal reasoning tasks.

Rank #17 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 86%
Percentile: 88.9%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: legal_bench; provider: OpenAI. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

88.9% percentile inside its fair comparison set

86%Raw benchmark valueCI 85.3% - 86.8%

Finance Agent v2

VALS-AI · Professional reasoning · Objective

Core financial analyst tasks for agentic models.

Rank #46 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 45.4%
Percentile: 34.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: fabv2; provider: OpenAI.

34.8% percentile inside its fair comparison set

45.4%Raw benchmark valueCI 44.5% - 46.2%

MedCode

VALS-AI · Professional reasoning · Objective

Medical billing support and coding tasks.

Rank #20 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 49.6%
Percentile: 81.1%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medcode; provider: OpenAI. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

81.1% percentile inside its fair comparison set

49.6%Raw benchmark valueCI 45.5% - 53.7%

MedScribe

VALS-AI · Professional reasoning · Objective

Administrative documentation support for doctors.

Rank #42 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 83.7%
Percentile: 58.3%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medscribe; provider: OpenAI. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

58.3% percentile inside its fair comparison set

83.7%Raw benchmark valueCI 79.9% - 87.4%

SAGE

VALS-AI · Professional reasoning · Objective

Student Assessment with Generative Evaluation.

Rank #15 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 50.8%
Percentile: 82.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: sage; provider: OpenAI.

82.7% percentile inside its fair comparison set

50.8%Raw benchmark valueCI 44.1% - 57.5%

TaxEval v2

VALS-AI · Professional reasoning · Objective

Answer quality on tax questions and responses.

Rank #78 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 71.2%
Percentile: 40.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: tax_eval_v2; provider: OpenAI.

40.3% percentile inside its fair comparison set

71.2%Raw benchmark valueCI 69.5% - 73%

Data analysis

LB · Professional reasoning · Objective

Structured data manipulation and table reasoning accuracy.

Rank #49 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 70.8%
Percentile: 26.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Data Analysis. Tasks scored: 3.

26.2% percentile inside its fair comparison set

70.8%Raw benchmark value

Overall

LB · Professional reasoning · Objective

Average objective performance across LiveBench's current public category mix.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 66.4%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category averages included: 7.

6.2% percentile inside its fair comparison set

66.4%Raw benchmark value

Consecutive events

LB · Professional reasoning · Objective

Objective consecutive events score in LiveBench.

Rank #53 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 63%
Percentile: 20%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: consecutive_events. Category: Data Analysis.

20% percentile inside its fair comparison set

63%Raw benchmark value

Table join

LB · Professional reasoning · Objective

Objective table join score in LiveBench.

Rank #25 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 49.4%
Percentile: 63.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: tablejoin. Category: Data Analysis.

63.1% percentile inside its fair comparison set

49.4%Raw benchmark value

Table reformat

LB · Professional reasoning · Objective

Objective table reformat score in LiveBench.

Rank #4 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 100%
Percentile: 100%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: tablereformat. Category: Data Analysis.

100% percentile inside its fair comparison set

100%Raw benchmark value

Hallucination

BB · Professional reasoning · Rubric

Accuracy and fabrication resistance under prompts that invite unsupported claims.

Rank #17 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 71.9%
Percentile: 52.9%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-hallucination.

52.9% percentile inside its fair comparison set

71.9%Raw benchmark value

BS pushback

BB · Professional reasoning · Rubric

Resistance to confidently accepting bogus assumptions in expert-style prompts.

Rank #9 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 78.5%
Percentile: 55.6%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-pushback.

55.6% percentile inside its fair comparison set

78.5%Raw benchmark value

Poker Agent

VALS-AI · Professional reasoning · Objective

Agent profit in poker-style strategic play.

Rank #4 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 1,103.2 score
Percentile: 94.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: poker_agent; provider: unknown. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

94.7% percentile inside its fair comparison set

1,103.2 scoreRaw benchmark valueCI 1,103.2 score - 1,103.2 score

Search / tool use3 benchmarks32.7%

Tau2-Bench Telecom

AA · Search / tool use · Objective

It matters when the model must browse, call tools, and recover useful answers from external systems.

Rank #229 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 23.4%
Percentile: 32.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `tau2`.

32.7% percentile inside its fair comparison set

23.4%Raw benchmark value

Search Arena

AR · Search / tool use · Human

It matters when the model must browse, call tools, and recover useful answers from external systems.

Rank #22 · Source label: gpt-5-search

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,183
Percentile: 39.4%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Arena leaderboard dataset row `gpt-5-search`. Category: overall. Source rank: #22. Votes: 20781. Organization: openai. License: Proprietary. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

39.4% percentile inside its fair comparison set

1,183Raw benchmark valueCI 1,175 - 1,190

Search Arena · No Style Control

AR · Search / tool use · Human

It matters when the model must browse, call tools, and recover useful answers from external systems.

Rank #29 · Source label: gpt-5-search

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,133
Percentile: 18.2%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Arena leaderboard dataset row `gpt-5-search`. Category: overall. Source rank: #30. Votes: 20781. Organization: openai. License: Proprietary. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

18.2% percentile inside its fair comparison set

1,133Raw benchmark valueCI 1,127 - 1,139

Long context2 benchmarks42.7%

Long Context Reasoning

AA · Long context · Objective

It checks whether long-context claims survive contact with retrieval, memory, or long-document tasks.

Rank #277 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 37%
Percentile: 39.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `lcr`.

39.3% percentile inside its fair comparison set

37%Raw benchmark value

CorpFin v2

VALS-AI · Long context · Objective

It checks whether long-context claims survive contact with retrieval, memory, or long-document tasks.

Rank #64 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 60.9%
Percentile: 46.2%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: corp_fin_v2; provider: OpenAI.

46.2% percentile inside its fair comparison set

60.9%Raw benchmark valueCI 59% - 62.8%

Vision understanding24 benchmarks46.3%

MMMU-Pro

AA · Vision understanding · Objective

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #151 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 58.4%
Percentile: 27.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `mmmuPro`.

27.9% percentile inside its fair comparison set

58.4%Raw benchmark value

VISTA

SL · Vision understanding · Rubric

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #5 · Source label: gpt-5-pro-2025-10-06

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 52.4%
Percentile: 95.2%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from the public Scale Labs page for scale-vista. Reported model configuration: gpt-5-pro-2025-10-06. Collapse policy: highest reported score per canonical model. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

95.2% percentile inside its fair comparison set

52.4%Raw benchmark value

VTB

SL · Vision understanding · Rubric

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #7 · Source label: gpt-5-2025-08-07-thinking

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 18.7%
Percentile: 79.2%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from the public Scale Labs page for scale-vtb. Reported model configuration: gpt-5-2025-08-07-thinking. Collapse policy: highest reported score per canonical model. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

79.2% percentile inside its fair comparison set

18.7%Raw benchmark value

Vision Arena

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #46 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,251
Percentile: 67.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #50. Votes: 25734. Organization: openai. License: Proprietary.

67.6% percentile inside its fair comparison set

1,251Raw benchmark valueCI 1,244 - 1,257

Vision Arena · Captioning

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #24 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,174
Percentile: 25.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: captioning. Source rank: #24. Votes: 283. Organization: openai. License: Proprietary.

25.8% percentile inside its fair comparison set

1,174Raw benchmark valueCI 1,131 - 1,217

Vision Arena · Creative Writing Vision

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #50 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,226
Percentile: 40.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing_vision. Source rank: #56. Votes: 1628. Organization: openai. License: Proprietary.

40.2% percentile inside its fair comparison set

1,226Raw benchmark valueCI 1,210 - 1,241

Vision Arena · Diagram

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #40 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,284
Percentile: 60.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: diagram. Source rank: #44. Votes: 6932. Organization: openai. License: Proprietary.

60.6% percentile inside its fair comparison set

1,284Raw benchmark valueCI 1,275 - 1,293

Vision Arena · English

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #47 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,249
Percentile: 66.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #52. Votes: 10678. Organization: openai. License: Proprietary.

66.7% percentile inside its fair comparison set

1,249Raw benchmark valueCI 1,240 - 1,257

Vision Arena · Entity Recognition

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #24 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,211
Percentile: 42.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: entity_recognition. Source rank: #27. Votes: 282. Organization: openai. License: Proprietary.

42.5% percentile inside its fair comparison set

1,211Raw benchmark valueCI 1,170 - 1,251

Vision Arena · Homework

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #36 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,291
Percentile: 63.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: homework. Source rank: #40. Votes: 3508. Organization: openai. License: Proprietary.

63.5% percentile inside its fair comparison set

1,291Raw benchmark valueCI 1,280 - 1,302

Vision Arena · Humor

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #38 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,234
Percentile: 49.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: humor. Source rank: #43. Votes: 870. Organization: openai. License: Proprietary.

49.3% percentile inside its fair comparison set

1,234Raw benchmark valueCI 1,213 - 1,255

Vision Arena · Ocr

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #46 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,263
Percentile: 55%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: ocr. Source rank: #50. Votes: 18283. Organization: openai. License: Proprietary.

55% percentile inside its fair comparison set

1,263Raw benchmark valueCI 1,256 - 1,269

Vision Arena · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #60 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,245
Percentile: 57.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #66. Votes: 25734. Organization: openai. License: Proprietary.

57.6% percentile inside its fair comparison set

1,245Raw benchmark valueCI 1,239 - 1,251

Vision Arena · Captioning · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #17 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,216
Percentile: 48.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: captioning. Source rank: #18. Votes: 283. Organization: openai. License: Proprietary.

48.4% percentile inside its fair comparison set

1,216Raw benchmark valueCI 1,180 - 1,253

Vision Arena · Creative Writing Vision · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #59 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,222
Percentile: 29.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing_vision. Source rank: #67. Votes: 1628. Organization: openai. License: Proprietary.

29.3% percentile inside its fair comparison set

1,222Raw benchmark valueCI 1,206 - 1,237

Vision Arena · Diagram · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #53 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,266
Percentile: 47.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: diagram. Source rank: #57. Votes: 6932. Organization: openai. License: Proprietary.

47.5% percentile inside its fair comparison set

1,266Raw benchmark valueCI 1,257 - 1,275

Vision Arena · English · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #61 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,244
Percentile: 56.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #66. Votes: 10678. Organization: openai. License: Proprietary.

56.5% percentile inside its fair comparison set

1,244Raw benchmark valueCI 1,236 - 1,253

Vision Arena · Entity Recognition · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #24 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,226
Percentile: 42.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: entity_recognition. Source rank: #28. Votes: 282. Organization: openai. License: Proprietary.

42.5% percentile inside its fair comparison set

1,226Raw benchmark valueCI 1,190 - 1,261

Vision Arena · Homework · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #41 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 58.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: homework. Source rank: #45. Votes: 3508. Organization: openai. License: Proprietary.

58.3% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,275 - 1,297

Vision Arena · Humor · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #46 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,232
Percentile: 38.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: humor. Source rank: #51. Votes: 870. Organization: openai. License: Proprietary.

38.4% percentile inside its fair comparison set

1,232Raw benchmark valueCI 1,211 - 1,253

Vision Arena · Ocr · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #57 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,253
Percentile: 44%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: ocr. Source rank: #63. Votes: 18283. Organization: openai. License: Proprietary.

44% percentile inside its fair comparison set

1,253Raw benchmark valueCI 1,247 - 1,259

MMMU Pro

VALS-AI · Vision understanding · Objective

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #48 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 79.2%
Percentile: 43.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: mmmu; provider: OpenAI.

43.4% percentile inside its fair comparison set

79.2%Raw benchmark valueCI 77.3% - 81.2%

Vision Arena · Creative Writing

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #28 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,152
Percentile: 15.6%
Last updated: archived
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: creative_writing. Source rank: #29. Votes: 1108. Organization: openai. License: Proprietary.

15.6% percentile inside its fair comparison set

1,152Raw benchmark valueCI 1,132 - 1,173

Vision Arena · Creative Writing · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #25 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,181
Percentile: 25%
Last updated: archived
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: creative_writing. Source rank: #26. Votes: 1108. Organization: openai. License: Proprietary.

25% percentile inside its fair comparison set

1,181Raw benchmark valueCI 1,163 - 1,199

Document understanding3 benchmarks40.3%

MortgageTax

VALS-AI · Document understanding · Objective

It matters when the job is reading PDFs, tables, forms, or mixed-layout documents rather than plain chat.

Rank #50 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 63.5%
Percentile: 43%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: mortgage_tax; provider: OpenAI.

43% percentile inside its fair comparison set

63.5%Raw benchmark valueCI 61.7% - 65.3%

Vals Multimodal Index

VALS-AI · Document understanding · Combined

It matters when the job is reading PDFs, tables, forms, or mixed-layout documents rather than plain chat.

Rank #21 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 54
Percentile: 37.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: vals_multimodal_index; provider: OpenAI.

37.5% percentile inside its fair comparison set

54Raw benchmark valueCI 51 - 58

Multimodal mix

OC · Document understanding · Objective

It matters when the job is reading PDFs, tables, forms, or mixed-layout documents rather than plain chat.

Rank #4 · Source label: gpt-5

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: OpenCompass
Raw value: 75.4%
Percentile: 92.9%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

92.9% percentile inside its fair comparison set

75.4%Raw benchmark value

Safety1 benchmarkn/a

MASK

SL · Safety · Rubric

Whether a model stays honest instead of covertly optimizing against the user.

Rank #19 · Source label: gpt-5-pro-2025-10-06

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 86%
Percentile: 74.6%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from the public Scale Labs page for scale-mask. Reported model configuration: gpt-5-pro-2025-10-06. Collapse policy: highest reported score per canonical model. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

74.6% percentile inside its fair comparison set

86%Raw benchmark value

Embeddings / retrieval1 benchmarkn/a

Retrieval

MTEB · Embeddings / retrieval · Retrieval

It is one of the few direct signals for retrieval stacks, where embedding quality matters more than chat style.

Rank #3 · Source label: gpt-5

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: MTEB
Raw value: 58.8 ndcg
Percentile: 100%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Embedding endpoint score. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

100% percentile inside its fair comparison set

58.8 ndcgRaw benchmark value

Multilingual20 benchmarksn/a

Text Arena · Chinese

AR · Multilingual · Human

Observed user preference in Arena's Text Arena chinese leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,473
Percentile: 73%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #103. Votes: 3908. Organization: openai. License: Proprietary.

73% percentile inside its fair comparison set

1,473Raw benchmark valueCI 1,463 - 1,484

Text Arena · French

AR · Multilingual · Human

Observed user preference in Arena's Text Arena french leaderboard.

Rank #59 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,476
Percentile: 77.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: french. Source rank: #65. Votes: 2229. Organization: openai. License: Proprietary.

77.4% percentile inside its fair comparison set

1,476Raw benchmark valueCI 1,461 - 1,491

Text Arena · German

AR · Multilingual · Human

Observed user preference in Arena's Text Arena german leaderboard.

Rank #70 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,442
Percentile: 74.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: german. Source rank: #82. Votes: 923. Organization: openai. License: Proprietary.

74.6% percentile inside its fair comparison set

1,442Raw benchmark valueCI 1,422 - 1,462

Text Arena · Japanese

AR · Multilingual · Human

Observed user preference in Arena's Text Arena japanese leaderboard.

Rank #66 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,409
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: japanese. Source rank: #79. Votes: 617. Organization: openai. License: Proprietary.

73.3% percentile inside its fair comparison set

1,409Raw benchmark valueCI 1,383 - 1,434

Text Arena · Korean

AR · Multilingual · Human

Observed user preference in Arena's Text Arena korean leaderboard.

Rank #75 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,393
Percentile: 69.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: korean. Source rank: #84. Votes: 987. Organization: openai. License: Proprietary.

69.7% percentile inside its fair comparison set

1,393Raw benchmark valueCI 1,373 - 1,413

Text Arena · Non English

AR · Multilingual · Human

Observed user preference in Arena's Text Arena non_english leaderboard.

Rank #72 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 81.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: non_english. Source rank: #79. Votes: 34227. Organization: openai. License: Proprietary.

81.1% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,433 - 1,442

Text Arena · Polish

AR · Multilingual · Human

Observed user preference in Arena's Text Arena polish leaderboard.

Rank #42 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,462
Percentile: 79.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: polish. Source rank: #49. Votes: 1196. Organization: openai. License: Proprietary.

79.4% percentile inside its fair comparison set

1,462Raw benchmark valueCI 1,445 - 1,480

Text Arena · Russian

AR · Multilingual · Human

Observed user preference in Arena's Text Arena russian leaderboard.

Rank #69 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,451
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: russian. Source rank: #75. Votes: 6979. Organization: openai. License: Proprietary.

80.1% percentile inside its fair comparison set

1,451Raw benchmark valueCI 1,443 - 1,459

Text Arena · Spanish

AR · Multilingual · Human

Observed user preference in Arena's Text Arena spanish leaderboard.

Rank #79 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,435
Percentile: 69.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: spanish. Source rank: #88. Votes: 1910. Organization: openai. License: Proprietary.

69.9% percentile inside its fair comparison set

1,435Raw benchmark valueCI 1,420 - 1,450

Text Arena · Chinese · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena chinese leaderboard.

Rank #122 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 65.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #133. Votes: 3908. Organization: openai. License: Proprietary.

65.6% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,435 - 1,456

Text Arena · French · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena french leaderboard.

Rank #88 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,440
Percentile: 66.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: french. Source rank: #95. Votes: 2229. Organization: openai. License: Proprietary.

66.1% percentile inside its fair comparison set

1,440Raw benchmark valueCI 1,425 - 1,454

Text Arena · German · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena german leaderboard.

Rank #97 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,407
Percentile: 64.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: german. Source rank: #109. Votes: 923. Organization: openai. License: Proprietary.

64.7% percentile inside its fair comparison set

1,407Raw benchmark valueCI 1,387 - 1,428

Text Arena · Japanese · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena japanese leaderboard.

Rank #88 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,375
Percentile: 64.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: japanese. Source rank: #100. Votes: 617. Organization: openai. License: Proprietary.

64.2% percentile inside its fair comparison set

1,375Raw benchmark valueCI 1,350 - 1,401

Text Arena · Korean · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena korean leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,368
Percentile: 61.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: korean. Source rank: #108. Votes: 987. Organization: openai. License: Proprietary.

61.1% percentile inside its fair comparison set

1,368Raw benchmark valueCI 1,348 - 1,388

Text Arena · Non English · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena non_english leaderboard.

Rank #111 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,404
Percentile: 70.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: non_english. Source rank: #121. Votes: 34227. Organization: openai. License: Proprietary.

70.7% percentile inside its fair comparison set

1,404Raw benchmark valueCI 1,400 - 1,409

Text Arena · Polish · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena polish leaderboard.

Rank #90 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,415
Percentile: 55.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: polish. Source rank: #100. Votes: 1196. Organization: openai. License: Proprietary.

55.3% percentile inside its fair comparison set

1,415Raw benchmark valueCI 1,398 - 1,433

Text Arena · Russian · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena russian leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,417
Percentile: 72.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: russian. Source rank: #105. Votes: 6979. Organization: openai. License: Proprietary.

72.1% percentile inside its fair comparison set

1,417Raw benchmark valueCI 1,409 - 1,425

Text Arena · Spanish · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena spanish leaderboard.

Rank #119 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,405
Percentile: 54.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: spanish. Source rank: #130. Votes: 1910. Organization: openai. License: Proprietary.

54.4% percentile inside its fair comparison set

1,405Raw benchmark valueCI 1,390 - 1,419

Vision Arena · Chinese

AR · Multilingual · Human

Observed user preference in Arena's Vision Arena chinese leaderboard.

Rank #45 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 57.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #50. Votes: 1659. Organization: openai. License: Proprietary.

57.7% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,268 - 1,303

Vision Arena · Chinese · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Vision Arena chinese leaderboard.

Rank #51 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 51.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #56. Votes: 1659. Organization: openai. License: Proprietary.

51.9% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,269 - 1,303

Source links and registry checks

official

Terminal-Bench

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Artificial Analysis

Oct 3, 2026

source →

official

LiveBench

Oct 3, 2026

source →

Model profile · OpenAI

GPT-5.4 mini

Closed weightspremium · registry tag 2026 mini

Verified but aging

Reads as verified but aging across the resolved source data. Capability fit excludes speed and cost.

Visible coverage: 65.8%
Verified coverage: 56.8%
Spread: 98.5%
Last verified: Oct 3, 2026

44%capability fit

textcodevisiondocumentsearch18 aliases46 official source links

Open compare

Data version

Current snapshot.

Published Oct 3, 2026Model list checked4 stale sources105 claim warnings9 providers · 1334 tracked models

The registry snapshot and page stamp are shown so a stale deploy is visible at a glance. View live data status.

Source-linked scores by benchmark

Verified but agingThis model currently reads as verified but aging across the resolved source data.

Chat / text36 benchmarks43.4%

Intelligence Index

AA · Chat / text · Combined

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #298 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 10
Percentile: 46.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `intelligenceIndex`.

46.5% percentile inside its fair comparison set

10Raw benchmark value

AA-Omniscience accuracy

AA · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #235 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 19.2%
Percentile: 48.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceAccuracy`.

48.3% percentile inside its fair comparison set

19.2%Raw benchmark value

AA-Omniscience non-hallucination

AA · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #430 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 4.4%
Percentile: 5.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceNonHallucination`.

5.3% percentile inside its fair comparison set

4.4%Raw benchmark value

IFBench

AA · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #205 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 38.8%
Percentile: 41.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `ifbench`.

41.4% percentile inside its fair comparison set

38.8%Raw benchmark value

Input price

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #209 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $0.8 /1M input tokens
Percentile: 41.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mInputTokens`.

41.2% percentile inside its fair comparison set

$0.8 /1M input tokensRaw benchmark value

Output price

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #243 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $4.5 /1M output tokens
Percentile: 31.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mOutputTokens`.

31.6% percentile inside its fair comparison set

$4.5 /1M output tokensRaw benchmark value

Output Speed

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #75 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 127.1 tokens/s
Percentile: 71.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianOutputTokensPerSecond`.

71.8% percentile inside its fair comparison set

127.1 tokens/sRaw benchmark value

Time to first token

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #253 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 123.46s
Percentile: 4.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstTokenSeconds`.

4.5% percentile inside its fair comparison set

123.46sRaw benchmark value

Time to first answer token

AA · Chat / text · Speed / cost

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #251 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 123.46s
Percentile: 4.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstAnswerTokenSeconds`.

4.6% percentile inside its fair comparison set

123.46sRaw benchmark value

Text Arena

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #76 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,447
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #84. Votes: 63290. Organization: openai. License: Proprietary.

80.1% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,443 - 1,451

Text Arena · Creative Writing

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #97 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,401
Percentile: 74.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing. Source rank: #107. Votes: 10797. Organization: openai. License: Proprietary.

74.3% percentile inside its fair comparison set

1,401Raw benchmark valueCI 1,394 - 1,409

Text Arena · English

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,450
Percentile: 77.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #92. Votes: 29062. Organization: openai. License: Proprietary.

77.9% percentile inside its fair comparison set

1,450Raw benchmark valueCI 1,445 - 1,455

Text Arena · Exclude Ties

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #76 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,449
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: exclude_ties. Source rank: #84. Votes: 48664. Organization: openai. License: Proprietary.

80.1% percentile inside its fair comparison set

1,449Raw benchmark valueCI 1,444 - 1,453

Text Arena · Hard Prompts

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #75 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,469
Percentile: 80.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts. Source rank: #83. Votes: 41568. Organization: openai. License: Proprietary.

80.3% percentile inside its fair comparison set

1,469Raw benchmark valueCI 1,464 - 1,473

Text Arena · Hard Prompts English

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #87 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,466
Percentile: 77%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts_english. Source rank: #95. Votes: 19794. Organization: openai. License: Proprietary.

77% percentile inside its fair comparison set

1,466Raw benchmark valueCI 1,460 - 1,472

Text Arena · Instruction Following

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #82 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,434
Percentile: 78.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: instruction_following. Source rank: #89. Votes: 21670. Organization: openai. License: Proprietary.

78.5% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,428 - 1,439

Text Arena · Longer Query

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #88 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,447
Percentile: 75.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: longer_query. Source rank: #96. Votes: 27722. Organization: openai. License: Proprietary.

75.5% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,441 - 1,452

Text Arena · Multi Turn

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #61 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,466
Percentile: 84%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: multi_turn. Source rank: #68. Votes: 11873. Organization: openai. License: Proprietary.

84% percentile inside its fair comparison set

1,466Raw benchmark valueCI 1,459 - 1,472

Text Arena · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #123 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,411
Percentile: 67.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #135. Votes: 63290. Organization: openai. License: Proprietary.

67.6% percentile inside its fair comparison set

1,411Raw benchmark valueCI 1,408 - 1,415

Text Arena · Creative Writing · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #126 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,370
Percentile: 66.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing. Source rank: #141. Votes: 10797. Organization: openai. License: Proprietary.

66.6% percentile inside its fair comparison set

1,370Raw benchmark valueCI 1,363 - 1,377

Text Arena · English · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #136 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,412
Percentile: 64.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #150. Votes: 29062. Organization: openai. License: Proprietary.

64.1% percentile inside its fair comparison set

1,412Raw benchmark valueCI 1,407 - 1,417

Text Arena · Exclude Ties · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #123 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,397
Percentile: 67.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: exclude_ties. Source rank: #134. Votes: 48664. Organization: openai. License: Proprietary.

67.6% percentile inside its fair comparison set

1,397Raw benchmark valueCI 1,393 - 1,402

Text Arena · Hard Prompts · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #112 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,424
Percentile: 70.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts. Source rank: #123. Votes: 41568. Organization: openai. License: Proprietary.

70.5% percentile inside its fair comparison set

1,424Raw benchmark valueCI 1,420 - 1,429

Text Arena · Hard Prompts English · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #127 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,420
Percentile: 66.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: hard_prompts_english. Source rank: #139. Votes: 19794. Organization: openai. License: Proprietary.

66.3% percentile inside its fair comparison set

1,420Raw benchmark valueCI 1,415 - 1,426

Text Arena · Instruction Following · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #105 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,405
Percentile: 72.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: instruction_following. Source rank: #114. Votes: 21670. Organization: openai. License: Proprietary.

72.3% percentile inside its fair comparison set

1,405Raw benchmark valueCI 1,399 - 1,410

Text Arena · Longer Query · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #121 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,407
Percentile: 66.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: longer_query. Source rank: #133. Votes: 27722. Organization: openai. License: Proprietary.

66.2% percentile inside its fair comparison set

1,407Raw benchmark valueCI 1,402 - 1,413

Text Arena · Multi Turn · No Style Control

AR · Chat / text · Human

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #92 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,429
Percentile: 75.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: multi_turn. Source rank: #102. Votes: 11873. Organization: openai. License: Proprietary.

75.7% percentile inside its fair comparison set

1,429Raw benchmark valueCI 1,422 - 1,436

Instruction following

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #59 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 59.8%
Percentile: 10.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: IF. Tasks scored: 4.

10.8% percentile inside its fair comparison set

59.8%Raw benchmark value

Language

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #61 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 71%
Percentile: 7.7%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Language. Tasks scored: 3.

7.7% percentile inside its fair comparison set

71%Raw benchmark value

Paraphrase

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #55 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 58.9%
Percentile: 16.9%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: paraphrase. Category: IF.

16.9% percentile inside its fair comparison set

58.9%Raw benchmark value

Simplify

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #60 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 54.1%
Percentile: 9.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: simplify. Category: IF.

9.2% percentile inside its fair comparison set

54.1%Raw benchmark value

Story generation

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #55 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 66.6%
Percentile: 16.9%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: story_generation. Category: IF.

16.9% percentile inside its fair comparison set

66.6%Raw benchmark value

Summarize

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #58 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 59.6%
Percentile: 12.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: summarize. Category: IF.

12.3% percentile inside its fair comparison set

59.6%Raw benchmark value

Connections

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #33 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 99%
Percentile: 50.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: connections. Category: Language.

50.8% percentile inside its fair comparison set

99%Raw benchmark value

Plot unscrambling

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 43.9%
Percentile: 3.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: plot_unscrambling. Category: Language.

3.1% percentile inside its fair comparison set

43.9%Raw benchmark value

Typos

LB · Chat / text · Objective

It tests whether the model is actually useful in normal conversational turns, not just on narrow correctness tasks.

Rank #60 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 70%
Percentile: 10.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: typos. Category: Language.

10.8% percentile inside its fair comparison set

70%Raw benchmark value

Coding30 benchmarks44.8%

Terminal-Bench Hard

AA · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #135 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 14.4%
Percentile: 59.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `terminalbenchHard`.

59.6% percentile inside its fair comparison set

14.4%Raw benchmark value

SciCode

AA · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #147 · Source label: GPT-5 mini (High)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 39%
Percentile: 26.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `scicode`.

26.3% percentile inside its fair comparison set

39%Raw benchmark value

Text Arena · Coding

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #79 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,495
Percentile: 79%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: coding. Source rank: #85. Votes: 17210. Organization: openai. License: Proprietary.

79% percentile inside its fair comparison set

1,495Raw benchmark valueCI 1,489 - 1,501

Text Arena · Coding · No Style Control

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #110 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,438
Percentile: 70.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: coding. Source rank: #119. Votes: 17210. Organization: openai. License: Proprietary.

70.6% percentile inside its fair comparison set

1,438Raw benchmark valueCI 1,432 - 1,444

Vibe Code Bench v1.1

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #46 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 48%
Percentile: 51.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: vibe-code; provider: OpenAI.

51.6% percentile inside its fair comparison set

48%Raw benchmark valueCI 37% - 59%

WebDev Arena

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,397
Percentile: 34.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #87. Votes: 11998. Organization: openai. License: Proprietary.

34.6% percentile inside its fair comparison set

1,397Raw benchmark valueCI 1,391 - 1,404

Code Arena · Webdev Html

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #76 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,420
Percentile: 40%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: webdev-html. Source rank: #81. Votes: 1437. Organization: openai. License: Proprietary.

40% percentile inside its fair comparison set

1,420Raw benchmark valueCI 1,404 - 1,437

Code Arena · Webdev React

AR · Coding · Human

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #81 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,386
Percentile: 29.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: webdev-react. Source rank: #86. Votes: 9878. Organization: openai. License: Proprietary.

29.2% percentile inside its fair comparison set

1,386Raw benchmark valueCI 1,379 - 1,393

Code Migration

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #54 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 12.9%
Percentile: 22.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: code-migration; provider: OpenAI.

22.1% percentile inside its fair comparison set

12.9%Raw benchmark valueCI 5.8% - 20.1%

ProgramBench

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #39 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 0%
Percentile: 51.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: programbench; provider: OpenAI.

51.7% percentile inside its fair comparison set

0%Raw benchmark valueCI 0% - 0%

Terminal-Bench 2.1

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #45 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 54.7%
Percentile: 34.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: terminal-bench-2-1; provider: OpenAI.

34.3% percentile inside its fair comparison set

54.7%Raw benchmark valueCI 45.8% - 63.6%

SWE-bench Verified

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #49 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 73%
Percentile: 41.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: swebench; provider: OpenAI.

41.5% percentile inside its fair comparison set

73%Raw benchmark valueCI 69.1% - 76.9%

LiveCodeBench

VALS-AI · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #59 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 81.5%
Percentile: 53.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: lcb; provider: OpenAI.

53.6% percentile inside its fair comparison set

81.5%Raw benchmark valueCI 79.3% - 83.6%

Agentic coding

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #58 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 41.7%
Percentile: 12.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Agentic Coding. Tasks scored: 3.

12.3% percentile inside its fair comparison set

41.7%Raw benchmark value

Coding

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #56 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 71.6%
Percentile: 15.4%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Coding. Tasks scored: 2.

15.4% percentile inside its fair comparison set

71.6%Raw benchmark value

JavaScript

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 50%
Percentile: 3.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: javascript. Category: Agentic Coding.

3.1% percentile inside its fair comparison set

50%Raw benchmark value

TypeScript

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #63 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 20%
Percentile: 7.7%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: typescript. Category: Agentic Coding.

7.7% percentile inside its fair comparison set

20%Raw benchmark value

Python

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #27 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 55%
Percentile: 63.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: python. Category: Agentic Coding.

63.1% percentile inside its fair comparison set

55%Raw benchmark value

Coding generation

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 67.7%
Percentile: 3.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: code_generation. Category: Coding.

3.1% percentile inside its fair comparison set

67.7%Raw benchmark value

Coding completion

LB · Coding · Objective

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #49 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 75.6%
Percentile: 26.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: code_completion. Category: Coding.

26.2% percentile inside its fair comparison set

75.6%Raw benchmark value

Debugging

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #14 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 84.1%
Percentile: 50%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-debugging.

50% percentile inside its fair comparison set

84.1%Raw benchmark value

Refactoring

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #13 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 62.3%
Percentile: 42.9%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-refactoring.

42.9% percentile inside its fair comparison set

62.3%Raw benchmark value

Security

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #5 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 83.3%
Percentile: 84%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-security.

84% percentile inside its fair comparison set

83.3%Raw benchmark value

UI bench

BB · Coding · Rubric

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #9 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 72.6%
Percentile: 63.6%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ui.

63.6% percentile inside its fair comparison set

72.6%Raw benchmark value

Speed throughput

BB · Coding · Speed / cost

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #4 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 236.4 t/s
Percentile: 88%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-throughput.

88% percentile inside its fair comparison set

236.4 t/sRaw benchmark value

Speed TTFT

BB · Coding · Speed / cost

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #1 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 233.00ms
Percentile: 100%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ttft.

100% percentile inside its fair comparison set

233.00msRaw benchmark value

Speed cost

BB · Coding · Speed / cost

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #7 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: $0.1 $
Percentile: 76%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-speed-cost.

76% percentile inside its fair comparison set

$0.1 $Raw benchmark value

Overall quality

BB · Coding · Combined

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #11 · Source label: GPT-5.4 Mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 77.6%
Percentile: 54.5%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-overall-quality.

54.5% percentile inside its fair comparison set

77.6%Raw benchmark value

Overall vibe

BB · Coding · Combined

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #11 · Source label: GPT-5.4 Mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 77.5%
Percentile: 54.5%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-overall-vibe.

54.5% percentile inside its fair comparison set

77.5%Raw benchmark value

Overall coverage

BB · Coding · Combined

It tells you whether the model can generate, repair, and reason over code under evaluator pressure rather than marketing examples.

Rank #9 · Source label: GPT-5.4 Mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 100%
Percentile: 100%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-overall-coverage.

100% percentile inside its fair comparison set

100%Raw benchmark value

Reasoning / math / science20 benchmarks32.5%

Humanity's Last Exam

AA · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #356 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 5.1%
Percentile: 32.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `hle`.

32.9% percentile inside its fair comparison set

5.1%Raw benchmark value

GPQA

AA · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #293 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 60.6%
Percentile: 40.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gpqa`.

40.8% percentile inside its fair comparison set

60.6%Raw benchmark value

CritPt

AA · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #229 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 0%
Percentile: 50.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `critpt`.

50.1% percentile inside its fair comparison set

0%Raw benchmark value

TutorBench

SL · Reasoning / math / science · Rubric

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #5 · Source label: gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 55.3%
Percentile: 88.9%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

88.9% percentile inside its fair comparison set

55.3%Raw benchmark value

MultiNRC

SL · Reasoning / math / science · Rubric

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #3 · Source label: gpt-5-pro-2025-10-06

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 65.2%
Percentile: 97.7%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

97.7% percentile inside its fair comparison set

65.2%Raw benchmark value

Text Arena · Math

AR · Reasoning / math / science · Human

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,439
Percentile: 76.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: math. Source rank: #91. Votes: 3361. Organization: openai. License: Proprietary.

76.9% percentile inside its fair comparison set

1,439Raw benchmark valueCI 1,428 - 1,450

Text Arena · Math · No Style Control

AR · Reasoning / math / science · Human

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #115 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,418
Percentile: 68.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: math. Source rank: #124. Votes: 3361. Organization: openai. License: Proprietary.

68.2% percentile inside its fair comparison set

1,418Raw benchmark valueCI 1,407 - 1,429

GPQA Diamond

VALS-AI · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #59 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 83.1%
Percentile: 53.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: gpqa; provider: OpenAI.

53.2% percentile inside its fair comparison set

83.1%Raw benchmark valueCI 78.3% - 87.9%

MMLU Pro

VALS-AI · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #63 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 84.6%
Percentile: 50.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: mmlu_pro; provider: OpenAI.

50.4% percentile inside its fair comparison set

84.6%Raw benchmark valueCI 83.9% - 85.3%

Mathematics

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 78.5%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Mathematics. Tasks scored: 4.

6.2% percentile inside its fair comparison set

78.5%Raw benchmark value

Reasoning

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 71.3%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Reasoning. Tasks scored: 4.

6.2% percentile inside its fair comparison set

71.3%Raw benchmark value

AMPS Hard

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #47 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 97%
Percentile: 32.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: AMPS_Hard. Category: Mathematics.

32.3% percentile inside its fair comparison set

97%Raw benchmark value

Integrals with game

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #64 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 37%
Percentile: 4.6%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: integrals_with_game. Category: Mathematics.

4.6% percentile inside its fair comparison set

37%Raw benchmark value

Math competition

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #50 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 94.1%
Percentile: 26.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: math_comp. Category: Mathematics.

26.2% percentile inside its fair comparison set

94.1%Raw benchmark value

Olympiad

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #58 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 85.7%
Percentile: 12.3%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: olympiad. Category: Mathematics.

12.3% percentile inside its fair comparison set

85.7%Raw benchmark value

Theory of mind

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #44 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 76.9%
Percentile: 33.8%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: theory_of_mind. Category: Reasoning.

33.8% percentile inside its fair comparison set

76.9%Raw benchmark value

Zebra puzzle

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 60.4%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: zebra_puzzle. Category: Reasoning.

6.2% percentile inside its fair comparison set

60.4%Raw benchmark value

Spatial

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #65 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 88%
Percentile: 1.5%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: spatial. Category: Reasoning.

1.5% percentile inside its fair comparison set

88%Raw benchmark value

Logic with navigation

LB · Reasoning / math / science · Objective

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #65 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 60%
Percentile: 1.5%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: logic_with_navigation. Category: Reasoning.

1.5% percentile inside its fair comparison set

60%Raw benchmark value

Reasoning

BB · Reasoning / math / science · Rubric

It is one of the cleaner reads on deliberate reasoning strength rather than style or popularity.

Rank #17 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 36.2%
Percentile: 23.8%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-reasoning.

23.8% percentile inside its fair comparison set

36.2%Raw benchmark value

Professional reasoning37 benchmarks40.6%

APEX-Agents-AA

AA · Professional reasoning · Objective

Long-horizon agentic task completion.

Rank #14 · Source label: GPT-5.4 mini (Xhigh)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 28.2%
Percentile: 59.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `apexAgents`.

59.4% percentile inside its fair comparison set

28.2%Raw benchmark value

PRBench Legal

SL · Professional reasoning · Rubric

Applied legal reasoning on professional-domain tasks.

Rank #9 · Source label: gpt-5-pro

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 49.9%
Percentile: 82.9%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

82.9% percentile inside its fair comparison set

49.9%Raw benchmark value

Text Arena · Expert

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena expert leaderboard.

Rank #69 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,480
Percentile: 79.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: expert. Source rank: #72. Votes: 6299. Organization: openai. License: Proprietary.

79.2% percentile inside its fair comparison set

1,480Raw benchmark valueCI 1,472 - 1,489

Text Arena · Industry Business And Management And Financial Operations

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_business_and_management_and_financial_operations leaderboard.

Rank #57 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,460
Percentile: 84.8%
Last updated: recent
Eligibility: headline eligible

84.8% percentile inside its fair comparison set

1,460Raw benchmark valueCI 1,453 - 1,466

Text Arena · Industry Entertainment And Sports And Media

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_entertainment_and_sports_and_media leaderboard.

Rank #84 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,409
Percentile: 77.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_entertainment_and_sports_and_media. Source rank: #92. Votes: 13880. Organization: openai. License: Proprietary.

77.8% percentile inside its fair comparison set

1,409Raw benchmark valueCI 1,402 - 1,415

Text Arena · Industry Legal And Government

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_legal_and_government leaderboard.

Rank #72 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,457
Percentile: 79.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_legal_and_government. Source rank: #82. Votes: 5040. Organization: openai. License: Proprietary.

79.7% percentile inside its fair comparison set

1,457Raw benchmark valueCI 1,448 - 1,466

Text Arena · Industry Life And Physical And Social Science

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_life_and_physical_and_social_science leaderboard.

Rank #77 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,465
Percentile: 79.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_life_and_physical_and_social_science. Source rank: #84. Votes: 10493. Organization: openai. License: Proprietary.

79.7% percentile inside its fair comparison set

1,465Raw benchmark valueCI 1,458 - 1,472

Text Arena · Industry Mathematical

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_mathematical leaderboard.

Rank #75 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,452
Percentile: 79.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_mathematical. Source rank: #81. Votes: 3420. Organization: openai. License: Proprietary.

79.2% percentile inside its fair comparison set

1,452Raw benchmark valueCI 1,441 - 1,463

Text Arena · Industry Medicine And Healthcare

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_medicine_and_healthcare leaderboard.

Rank #95 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,451
Percentile: 72.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_medicine_and_healthcare. Source rank: #108. Votes: 4757. Organization: openai. License: Proprietary.

72.8% percentile inside its fair comparison set

1,451Raw benchmark valueCI 1,441 - 1,460

Text Arena · Industry Software And It Services

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_software_and_it_services leaderboard.

Rank #77 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,485
Percentile: 79.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_software_and_it_services. Source rank: #84. Votes: 24993. Organization: openai. License: Proprietary.

79.8% percentile inside its fair comparison set

1,485Raw benchmark valueCI 1,480 - 1,490

Text Arena · Industry Writing And Literature And Language

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_writing_and_literature_and_language leaderboard.

Rank #79 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,423
Percentile: 79.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_writing_and_literature_and_language. Source rank: #88. Votes: 15820. Organization: openai. License: Proprietary.

79.2% percentile inside its fair comparison set

1,423Raw benchmark valueCI 1,417 - 1,429

Text Arena · Expert · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena expert leaderboard.

Rank #99 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,435
Percentile: 70%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: expert. Source rank: #105. Votes: 6299. Organization: openai. License: Proprietary.

70% percentile inside its fair comparison set

1,435Raw benchmark valueCI 1,426 - 1,443

Text Arena · Industry Business And Management And Financial Operations · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_business_and_management_and_financial_operations leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,418
Percentile: 74.3%
Last updated: recent
Eligibility: headline eligible

74.3% percentile inside its fair comparison set

1,418Raw benchmark valueCI 1,411 - 1,424

Text Arena · Industry Entertainment And Sports And Media · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_entertainment_and_sports_and_media leaderboard.

Rank #124 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,373
Percentile: 67.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_entertainment_and_sports_and_media. Source rank: #136. Votes: 13880. Organization: openai. License: Proprietary.

67.1% percentile inside its fair comparison set

1,373Raw benchmark valueCI 1,366 - 1,379

Text Arena · Industry Legal And Government · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_legal_and_government leaderboard.

Rank #128 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,415
Percentile: 63.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_legal_and_government. Source rank: #141. Votes: 5040. Organization: openai. License: Proprietary.

63.6% percentile inside its fair comparison set

1,415Raw benchmark valueCI 1,406 - 1,424

Text Arena · Industry Life And Physical And Social Science · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_life_and_physical_and_social_science leaderboard.

Rank #131 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,417
Percentile: 65.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_life_and_physical_and_social_science. Source rank: #142. Votes: 10493. Organization: openai. License: Proprietary.

65.2% percentile inside its fair comparison set

1,417Raw benchmark valueCI 1,410 - 1,424

Text Arena · Industry Mathematical · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_mathematical leaderboard.

Rank #110 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,428
Percentile: 69.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_mathematical. Source rank: #117. Votes: 3420. Organization: openai. License: Proprietary.

69.4% percentile inside its fair comparison set

1,428Raw benchmark valueCI 1,417 - 1,439

Text Arena · Industry Medicine And Healthcare · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_medicine_and_healthcare leaderboard.

Rank #143 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,395
Percentile: 59%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_medicine_and_healthcare. Source rank: #159. Votes: 4757. Organization: openai. License: Proprietary.

59% percentile inside its fair comparison set

1,395Raw benchmark valueCI 1,385 - 1,404

Text Arena · Industry Software And It Services · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_software_and_it_services leaderboard.

Rank #115 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,436
Percentile: 69.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_software_and_it_services. Source rank: #125. Votes: 24993. Organization: openai. License: Proprietary.

69.7% percentile inside its fair comparison set

1,436Raw benchmark valueCI 1,430 - 1,441

Text Arena · Industry Writing And Literature And Language · No Style Control

AR · Professional reasoning · Human

Observed user preference in Arena's Text Arena industry_writing_and_literature_and_language leaderboard.

Rank #113 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,392
Percentile: 70.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: industry_writing_and_literature_and_language. Source rank: #123. Votes: 15820. Organization: openai. License: Proprietary.

70.1% percentile inside its fair comparison set

1,392Raw benchmark valueCI 1,386 - 1,398

Vals Index

VALS-AI · Professional reasoning · Combined

Weighted model performance across economically relevant Vals tasks.

Rank #38 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 33
Percentile: 7.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: vals_index; provider: OpenAI.

7.5% percentile inside its fair comparison set

33Raw benchmark valueCI 31 - 36

Legal Research Bench

VALS-AI · Professional reasoning · Objective

Applied legal research tasks.

Rank #59 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 12.5%
Percentile: 14.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: legal_research; provider: OpenAI.

14.7% percentile inside its fair comparison set

12.5%Raw benchmark valueCI 8% - 17%

Harvey's Legal Agent Benchmark

VALS-AI · Professional reasoning · Objective

Completing legal work with documents, spreadsheets, presentations, and file-system tools.

Rank #67 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 0%
Percentile: 20.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: hlab; provider: OpenAI.

20.3% percentile inside its fair comparison set

0%Raw benchmark valueCI 0% - 0%

LegalBench

VALS-AI · Professional reasoning · Objective

Academic legal reasoning tasks.

Rank #17 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 86%
Percentile: 88.9%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: legal_bench; provider: OpenAI. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

88.9% percentile inside its fair comparison set

86%Raw benchmark valueCI 85.3% - 86.8%

Finance Agent v2

VALS-AI · Professional reasoning · Objective

Core financial analyst tasks for agentic models.

Rank #46 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 45.4%
Percentile: 34.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: fabv2; provider: OpenAI.

34.8% percentile inside its fair comparison set

45.4%Raw benchmark valueCI 44.5% - 46.2%

MedCode

VALS-AI · Professional reasoning · Objective

Medical billing support and coding tasks.

Rank #20 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 49.6%
Percentile: 81.1%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medcode; provider: OpenAI. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

81.1% percentile inside its fair comparison set

49.6%Raw benchmark valueCI 45.5% - 53.7%

MedScribe

VALS-AI · Professional reasoning · Objective

Administrative documentation support for doctors.

Rank #42 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 83.7%
Percentile: 58.3%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medscribe; provider: OpenAI. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

58.3% percentile inside its fair comparison set

83.7%Raw benchmark valueCI 79.9% - 87.4%

SAGE

VALS-AI · Professional reasoning · Objective

Student Assessment with Generative Evaluation.

Rank #15 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 50.8%
Percentile: 82.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: sage; provider: OpenAI.

82.7% percentile inside its fair comparison set

50.8%Raw benchmark valueCI 44.1% - 57.5%

TaxEval v2

VALS-AI · Professional reasoning · Objective

Answer quality on tax questions and responses.

Rank #78 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 71.2%
Percentile: 40.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: tax_eval_v2; provider: OpenAI.

40.3% percentile inside its fair comparison set

71.2%Raw benchmark valueCI 69.5% - 73%

Data analysis

LB · Professional reasoning · Objective

Structured data manipulation and table reasoning accuracy.

Rank #49 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 70.8%
Percentile: 26.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category: Data Analysis. Tasks scored: 3.

26.2% percentile inside its fair comparison set

70.8%Raw benchmark value

Overall

LB · Professional reasoning · Objective

Average objective performance across LiveBench's current public category mix.

Rank #62 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 66.4%
Percentile: 6.2%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Category averages included: 7.

6.2% percentile inside its fair comparison set

66.4%Raw benchmark value

Consecutive events

LB · Professional reasoning · Objective

Objective consecutive events score in LiveBench.

Rank #53 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 63%
Percentile: 20%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: consecutive_events. Category: Data Analysis.

20% percentile inside its fair comparison set

63%Raw benchmark value

Table join

LB · Professional reasoning · Objective

Objective table join score in LiveBench.

Rank #25 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 49.4%
Percentile: 63.1%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: tablejoin. Category: Data Analysis.

63.1% percentile inside its fair comparison set

49.4%Raw benchmark value

Table reformat

LB · Professional reasoning · Objective

Objective table reformat score in LiveBench.

Rank #4 · Source label: gpt-5.4-mini-xhigh

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: LiveBench
Raw value: 100%
Percentile: 100%
Last updated: stale
Eligibility: headline eligible

Derived from the official LiveBench website leaderboard table. Task: tablereformat. Category: Data Analysis.

100% percentile inside its fair comparison set

100%Raw benchmark value

Hallucination

BB · Professional reasoning · Rubric

Accuracy and fabrication resistance under prompts that invite unsupported claims.

Rank #17 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 71.9%
Percentile: 52.9%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-hallucination.

52.9% percentile inside its fair comparison set

71.9%Raw benchmark value

BS pushback

BB · Professional reasoning · Rubric

Resistance to confidently accepting bogus assumptions in expert-style prompts.

Rank #9 · Source label: openai/gpt-5.4-mini

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 78.5%
Percentile: 55.6%
Last updated: archived
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-pushback.

55.6% percentile inside its fair comparison set

78.5%Raw benchmark value

Poker Agent

VALS-AI · Professional reasoning · Objective

Agent profit in poker-style strategic play.

Rank #4 · Source label: openai/gpt-5-2025-08-07

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 1,103.2 score
Percentile: 94.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: poker_agent; provider: unknown. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

94.7% percentile inside its fair comparison set

1,103.2 scoreRaw benchmark valueCI 1,103.2 score - 1,103.2 score

Search / tool use3 benchmarks32.7%

Tau2-Bench Telecom

AA · Search / tool use · Objective

It matters when the model must browse, call tools, and recover useful answers from external systems.

Rank #229 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 23.4%
Percentile: 32.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `tau2`.

32.7% percentile inside its fair comparison set

23.4%Raw benchmark value

Search Arena

AR · Search / tool use · Human

It matters when the model must browse, call tools, and recover useful answers from external systems.

Rank #22 · Source label: gpt-5-search

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,183
Percentile: 39.4%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

39.4% percentile inside its fair comparison set

1,183Raw benchmark valueCI 1,175 - 1,190

Search Arena · No Style Control

AR · Search / tool use · Human

It matters when the model must browse, call tools, and recover useful answers from external systems.

Rank #29 · Source label: gpt-5-search

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,133
Percentile: 18.2%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

18.2% percentile inside its fair comparison set

1,133Raw benchmark valueCI 1,127 - 1,139

Long context2 benchmarks42.7%

Long Context Reasoning

AA · Long context · Objective

It checks whether long-context claims survive contact with retrieval, memory, or long-document tasks.

Rank #277 · Source label: GPT-5.4 mini (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 37%
Percentile: 39.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `lcr`.

39.3% percentile inside its fair comparison set

37%Raw benchmark value

CorpFin v2

VALS-AI · Long context · Objective

It checks whether long-context claims survive contact with retrieval, memory, or long-document tasks.

Rank #64 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 60.9%
Percentile: 46.2%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: corp_fin_v2; provider: OpenAI.

46.2% percentile inside its fair comparison set

60.9%Raw benchmark valueCI 59% - 62.8%

Vision understanding24 benchmarks46.3%

MMMU-Pro

AA · Vision understanding · Objective

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #151 · Source label: GPT-5 mini (Minimal)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 58.4%
Percentile: 27.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `mmmuPro`.

27.9% percentile inside its fair comparison set

58.4%Raw benchmark value

VISTA

SL · Vision understanding · Rubric

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #5 · Source label: gpt-5-pro-2025-10-06

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 52.4%
Percentile: 95.2%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

95.2% percentile inside its fair comparison set

52.4%Raw benchmark value

VTB

SL · Vision understanding · Rubric

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #7 · Source label: gpt-5-2025-08-07-thinking

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 18.7%
Percentile: 79.2%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

79.2% percentile inside its fair comparison set

18.7%Raw benchmark value

Vision Arena

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #46 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,251
Percentile: 67.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #50. Votes: 25734. Organization: openai. License: Proprietary.

67.6% percentile inside its fair comparison set

1,251Raw benchmark valueCI 1,244 - 1,257

Vision Arena · Captioning

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #24 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,174
Percentile: 25.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: captioning. Source rank: #24. Votes: 283. Organization: openai. License: Proprietary.

25.8% percentile inside its fair comparison set

1,174Raw benchmark valueCI 1,131 - 1,217

Vision Arena · Creative Writing Vision

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #50 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,226
Percentile: 40.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing_vision. Source rank: #56. Votes: 1628. Organization: openai. License: Proprietary.

40.2% percentile inside its fair comparison set

1,226Raw benchmark valueCI 1,210 - 1,241

Vision Arena · Diagram

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #40 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,284
Percentile: 60.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: diagram. Source rank: #44. Votes: 6932. Organization: openai. License: Proprietary.

60.6% percentile inside its fair comparison set

1,284Raw benchmark valueCI 1,275 - 1,293

Vision Arena · English

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #47 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,249
Percentile: 66.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #52. Votes: 10678. Organization: openai. License: Proprietary.

66.7% percentile inside its fair comparison set

1,249Raw benchmark valueCI 1,240 - 1,257

Vision Arena · Entity Recognition

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #24 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,211
Percentile: 42.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: entity_recognition. Source rank: #27. Votes: 282. Organization: openai. License: Proprietary.

42.5% percentile inside its fair comparison set

1,211Raw benchmark valueCI 1,170 - 1,251

Vision Arena · Homework

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #36 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,291
Percentile: 63.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: homework. Source rank: #40. Votes: 3508. Organization: openai. License: Proprietary.

63.5% percentile inside its fair comparison set

1,291Raw benchmark valueCI 1,280 - 1,302

Vision Arena · Humor

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #38 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,234
Percentile: 49.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: humor. Source rank: #43. Votes: 870. Organization: openai. License: Proprietary.

49.3% percentile inside its fair comparison set

1,234Raw benchmark valueCI 1,213 - 1,255

Vision Arena · Ocr

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #46 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,263
Percentile: 55%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: ocr. Source rank: #50. Votes: 18283. Organization: openai. License: Proprietary.

55% percentile inside its fair comparison set

1,263Raw benchmark valueCI 1,256 - 1,269

Vision Arena · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #60 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,245
Percentile: 57.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: overall. Source rank: #66. Votes: 25734. Organization: openai. License: Proprietary.

57.6% percentile inside its fair comparison set

1,245Raw benchmark valueCI 1,239 - 1,251

Vision Arena · Captioning · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #17 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,216
Percentile: 48.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: captioning. Source rank: #18. Votes: 283. Organization: openai. License: Proprietary.

48.4% percentile inside its fair comparison set

1,216Raw benchmark valueCI 1,180 - 1,253

Vision Arena · Creative Writing Vision · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #59 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,222
Percentile: 29.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: creative_writing_vision. Source rank: #67. Votes: 1628. Organization: openai. License: Proprietary.

29.3% percentile inside its fair comparison set

1,222Raw benchmark valueCI 1,206 - 1,237

Vision Arena · Diagram · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #53 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,266
Percentile: 47.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: diagram. Source rank: #57. Votes: 6932. Organization: openai. License: Proprietary.

47.5% percentile inside its fair comparison set

1,266Raw benchmark valueCI 1,257 - 1,275

Vision Arena · English · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #61 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,244
Percentile: 56.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: english. Source rank: #66. Votes: 10678. Organization: openai. License: Proprietary.

56.5% percentile inside its fair comparison set

1,244Raw benchmark valueCI 1,236 - 1,253

Vision Arena · Entity Recognition · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #24 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,226
Percentile: 42.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: entity_recognition. Source rank: #28. Votes: 282. Organization: openai. License: Proprietary.

42.5% percentile inside its fair comparison set

1,226Raw benchmark valueCI 1,190 - 1,261

Vision Arena · Homework · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #41 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 58.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: homework. Source rank: #45. Votes: 3508. Organization: openai. License: Proprietary.

58.3% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,275 - 1,297

Vision Arena · Humor · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #46 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,232
Percentile: 38.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: humor. Source rank: #51. Votes: 870. Organization: openai. License: Proprietary.

38.4% percentile inside its fair comparison set

1,232Raw benchmark valueCI 1,211 - 1,253

Vision Arena · Ocr · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #57 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,253
Percentile: 44%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: ocr. Source rank: #63. Votes: 18283. Organization: openai. License: Proprietary.

44% percentile inside its fair comparison set

1,253Raw benchmark valueCI 1,247 - 1,259

MMMU Pro

VALS-AI · Vision understanding · Objective

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #48 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 79.2%
Percentile: 43.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: mmmu; provider: OpenAI.

43.4% percentile inside its fair comparison set

79.2%Raw benchmark valueCI 77.3% - 81.2%

Vision Arena · Creative Writing

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #28 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,152
Percentile: 15.6%
Last updated: archived
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: creative_writing. Source rank: #29. Votes: 1108. Organization: openai. License: Proprietary.

15.6% percentile inside its fair comparison set

1,152Raw benchmark valueCI 1,132 - 1,173

Vision Arena · Creative Writing · No Style Control

AR · Vision understanding · Human

It is useful when the model must read charts, UI, screenshots, or visual scenes rather than text alone.

Rank #25 · Source label: gpt-5-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,181
Percentile: 25%
Last updated: archived
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5-mini-high`. Category: creative_writing. Source rank: #26. Votes: 1108. Organization: openai. License: Proprietary.

25% percentile inside its fair comparison set

1,181Raw benchmark valueCI 1,163 - 1,199

Document understanding3 benchmarks40.3%

MortgageTax

VALS-AI · Document understanding · Objective

It matters when the job is reading PDFs, tables, forms, or mixed-layout documents rather than plain chat.

Rank #50 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 63.5%
Percentile: 43%
Last updated: recent
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: mortgage_tax; provider: OpenAI.

43% percentile inside its fair comparison set

63.5%Raw benchmark valueCI 61.7% - 65.3%

Vals Multimodal Index

VALS-AI · Document understanding · Combined

It matters when the job is reading PDFs, tables, forms, or mixed-layout documents rather than plain chat.

Rank #21 · Source label: openai/gpt-5.4-mini-2026-03-17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 54
Percentile: 37.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: vals_multimodal_index; provider: OpenAI.

37.5% percentile inside its fair comparison set

54Raw benchmark valueCI 51 - 58

Multimodal mix

OC · Document understanding · Objective

It matters when the job is reading PDFs, tables, forms, or mixed-layout documents rather than plain chat.

Rank #4 · Source label: gpt-5

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: OpenCompass
Raw value: 75.4%
Percentile: 92.9%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

92.9% percentile inside its fair comparison set

75.4%Raw benchmark value

Safety1 benchmarkn/a

MASK

SL · Safety · Rubric

Whether a model stays honest instead of covertly optimizing against the user.

Rank #19 · Source label: gpt-5-pro-2025-10-06

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 86%
Percentile: 74.6%
Last updated: recent
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

74.6% percentile inside its fair comparison set

86%Raw benchmark value

Embeddings / retrieval1 benchmarkn/a

Retrieval

MTEB · Embeddings / retrieval · Retrieval

It is one of the few direct signals for retrieval stacks, where embedding quality matters more than chat style.

Rank #3 · Source label: gpt-5

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: MTEB
Raw value: 58.8 ndcg
Percentile: 100%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Embedding endpoint score. Backfilled from GPT-5 via approved benchmark identity mapping map-gpt-5-4-mini-to-gpt-5.

100% percentile inside its fair comparison set

58.8 ndcgRaw benchmark value

Multilingual20 benchmarksn/a

Text Arena · Chinese

AR · Multilingual · Human

Observed user preference in Arena's Text Arena chinese leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,473
Percentile: 73%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #103. Votes: 3908. Organization: openai. License: Proprietary.

73% percentile inside its fair comparison set

1,473Raw benchmark valueCI 1,463 - 1,484

Text Arena · French

AR · Multilingual · Human

Observed user preference in Arena's Text Arena french leaderboard.

Rank #59 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,476
Percentile: 77.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: french. Source rank: #65. Votes: 2229. Organization: openai. License: Proprietary.

77.4% percentile inside its fair comparison set

1,476Raw benchmark valueCI 1,461 - 1,491

Text Arena · German

AR · Multilingual · Human

Observed user preference in Arena's Text Arena german leaderboard.

Rank #70 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,442
Percentile: 74.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: german. Source rank: #82. Votes: 923. Organization: openai. License: Proprietary.

74.6% percentile inside its fair comparison set

1,442Raw benchmark valueCI 1,422 - 1,462

Text Arena · Japanese

AR · Multilingual · Human

Observed user preference in Arena's Text Arena japanese leaderboard.

Rank #66 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,409
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: japanese. Source rank: #79. Votes: 617. Organization: openai. License: Proprietary.

73.3% percentile inside its fair comparison set

1,409Raw benchmark valueCI 1,383 - 1,434

Text Arena · Korean

AR · Multilingual · Human

Observed user preference in Arena's Text Arena korean leaderboard.

Rank #75 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,393
Percentile: 69.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: korean. Source rank: #84. Votes: 987. Organization: openai. License: Proprietary.

69.7% percentile inside its fair comparison set

1,393Raw benchmark valueCI 1,373 - 1,413

Text Arena · Non English

AR · Multilingual · Human

Observed user preference in Arena's Text Arena non_english leaderboard.

Rank #72 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 81.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: non_english. Source rank: #79. Votes: 34227. Organization: openai. License: Proprietary.

81.1% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,433 - 1,442

Text Arena · Polish

AR · Multilingual · Human

Observed user preference in Arena's Text Arena polish leaderboard.

Rank #42 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,462
Percentile: 79.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: polish. Source rank: #49. Votes: 1196. Organization: openai. License: Proprietary.

79.4% percentile inside its fair comparison set

1,462Raw benchmark valueCI 1,445 - 1,480

Text Arena · Russian

AR · Multilingual · Human

Observed user preference in Arena's Text Arena russian leaderboard.

Rank #69 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,451
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: russian. Source rank: #75. Votes: 6979. Organization: openai. License: Proprietary.

80.1% percentile inside its fair comparison set

1,451Raw benchmark valueCI 1,443 - 1,459

Text Arena · Spanish

AR · Multilingual · Human

Observed user preference in Arena's Text Arena spanish leaderboard.

Rank #79 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,435
Percentile: 69.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: spanish. Source rank: #88. Votes: 1910. Organization: openai. License: Proprietary.

69.9% percentile inside its fair comparison set

1,435Raw benchmark valueCI 1,420 - 1,450

Text Arena · Chinese · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena chinese leaderboard.

Rank #122 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 65.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #133. Votes: 3908. Organization: openai. License: Proprietary.

65.6% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,435 - 1,456

Text Arena · French · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena french leaderboard.

Rank #88 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,440
Percentile: 66.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: french. Source rank: #95. Votes: 2229. Organization: openai. License: Proprietary.

66.1% percentile inside its fair comparison set

1,440Raw benchmark valueCI 1,425 - 1,454

Text Arena · German · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena german leaderboard.

Rank #97 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,407
Percentile: 64.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: german. Source rank: #109. Votes: 923. Organization: openai. License: Proprietary.

64.7% percentile inside its fair comparison set

1,407Raw benchmark valueCI 1,387 - 1,428

Text Arena · Japanese · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena japanese leaderboard.

Rank #88 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,375
Percentile: 64.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: japanese. Source rank: #100. Votes: 617. Organization: openai. License: Proprietary.

64.2% percentile inside its fair comparison set

1,375Raw benchmark valueCI 1,350 - 1,401

Text Arena · Korean · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena korean leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,368
Percentile: 61.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: korean. Source rank: #108. Votes: 987. Organization: openai. License: Proprietary.

61.1% percentile inside its fair comparison set

1,368Raw benchmark valueCI 1,348 - 1,388

Text Arena · Non English · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena non_english leaderboard.

Rank #111 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,404
Percentile: 70.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: non_english. Source rank: #121. Votes: 34227. Organization: openai. License: Proprietary.

70.7% percentile inside its fair comparison set

1,404Raw benchmark valueCI 1,400 - 1,409

Text Arena · Polish · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena polish leaderboard.

Rank #90 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,415
Percentile: 55.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: polish. Source rank: #100. Votes: 1196. Organization: openai. License: Proprietary.

55.3% percentile inside its fair comparison set

1,415Raw benchmark valueCI 1,398 - 1,433

Text Arena · Russian · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena russian leaderboard.

Rank #96 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,417
Percentile: 72.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: russian. Source rank: #105. Votes: 6979. Organization: openai. License: Proprietary.

72.1% percentile inside its fair comparison set

1,417Raw benchmark valueCI 1,409 - 1,425

Text Arena · Spanish · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Text Arena spanish leaderboard.

Rank #119 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,405
Percentile: 54.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: spanish. Source rank: #130. Votes: 1910. Organization: openai. License: Proprietary.

54.4% percentile inside its fair comparison set

1,405Raw benchmark valueCI 1,390 - 1,419

Vision Arena · Chinese

AR · Multilingual · Human

Observed user preference in Arena's Vision Arena chinese leaderboard.

Rank #45 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 57.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #50. Votes: 1659. Organization: openai. License: Proprietary.

57.7% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,268 - 1,303

Vision Arena · Chinese · No Style Control

AR · Multilingual · Human

Observed user preference in Arena's Vision Arena chinese leaderboard.

Rank #51 · Source label: gpt-5.4-mini-high

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 51.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gpt-5.4-mini-high`. Category: chinese. Source rank: #56. Votes: 1659. Organization: openai. License: Proprietary.

51.9% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,269 - 1,303

Source links and registry checks

official

Terminal-Bench

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Arena

Oct 3, 2026

source →

official

Artificial Analysis

Oct 3, 2026

source →

official

LiveBench

Oct 3, 2026

source →

GPT-5.4 mini

Current snapshot.

Source-linked scores by benchmark

Source links and registry checks

Loading model evidence.

GPT-5.4 mini

Current snapshot.

Source-linked scores by benchmark

Source links and registry checks