Model profile · Google

Gemma 4 26B A4B

Open weightsmid · registry tag 2026 open multimodal

Thin verified coverage

Reads as thin verified coverage across the resolved source data. Capability fit excludes speed and cost.

Visible coverage: 13.7%
Verified coverage: 13.7%
Spread: 58.2%
Last verified: Oct 3, 2026

52%capability fit

textvisioncodedocument5 aliases39 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.

Thin verified coverageThis model currently reads as thin verified coverage across the resolved source data.

Chat / text27 benchmarks56.4%

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 #240 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 13
Percentile: 56.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `intelligenceIndex`.

56.9% percentile inside its fair comparison set

13Raw 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 #317 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 15.5%
Percentile: 30.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceAccuracy`.

30.2% percentile inside its fair comparison set

15.5%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 #373 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 7.9%
Percentile: 17.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceNonHallucination`.

17.9% percentile inside its fair comparison set

7.9%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 #137 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 45.4%
Percentile: 60.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `ifbench`.

60.9% percentile inside its fair comparison set

45.4%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 #70 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $0.1 /1M input tokens
Percentile: 80.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mInputTokens`.

80.5% percentile inside its fair comparison set

$0.1 /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 #68 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $0.4 /1M output tokens
Percentile: 81.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mOutputTokens`.

81.4% percentile inside its fair comparison set

$0.4 /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 #95 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 109.1 tokens/s
Percentile: 64.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianOutputTokensPerSecond`.

64.1% percentile inside its fair comparison set

109.1 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 #49 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 1.23s
Percentile: 81.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstTokenSeconds`.

81.8% percentile inside its fair comparison set

1.23sRaw 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 #29 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 1.23s
Percentile: 89.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstAnswerTokenSeconds`.

89.3% percentile inside its fair comparison set

1.23sRaw 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 #91

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 76.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #100. Votes: 6081. Organization: google. License: Apache 2.0.

76.1% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,430 - 1,445

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 #101

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,400
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing. Source rank: #111. Votes: 995. Organization: google. License: Apache 2.0.

73.3% percentile inside its fair comparison set

1,400Raw benchmark valueCI 1,381 - 1,418

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 #86

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,450
Percentile: 77.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #94. Votes: 2244. Organization: google. License: Apache 2.0.

77.4% percentile inside its fair comparison set

1,450Raw benchmark valueCI 1,438 - 1,462

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 #90

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,434
Percentile: 76.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: exclude_ties. Source rank: #99. Votes: 4155. Organization: google. License: Apache 2.0.

76.3% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,423 - 1,445

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 #88

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,461
Percentile: 76.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts. Source rank: #96. Votes: 3469. Organization: google. License: Apache 2.0.

76.9% percentile inside its fair comparison set

1,461Raw benchmark valueCI 1,451 - 1,470

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 #77

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,471
Percentile: 79.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts_english. Source rank: #84. Votes: 1272. Organization: google. License: Apache 2.0.

79.7% percentile inside its fair comparison set

1,471Raw benchmark valueCI 1,455 - 1,487

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 #76

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: instruction_following. Source rank: #83. Votes: 1737. Organization: google. License: Apache 2.0.

80.1% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,424 - 1,451

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 #89

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 75.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: longer_query. Source rank: #98. Votes: 1713. Organization: google. License: Apache 2.0.

75.2% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,432 - 1,460

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 #84

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,448
Percentile: 77.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: multi_turn. Source rank: #94. Votes: 1146. Organization: google. License: Apache 2.0.

77.8% percentile inside its fair comparison set

1,448Raw benchmark valueCI 1,431 - 1,465

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 #82

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 `gemma-4-26b-a4b`. Category: overall. Source rank: #89. Votes: 6081. Organization: google. License: Apache 2.0.

78.5% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,426 - 1,441

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 #89

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,401
Percentile: 76.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing. Source rank: #98. Votes: 995. Organization: google. License: Apache 2.0.

76.5% percentile inside its fair comparison set

1,401Raw benchmark valueCI 1,383 - 1,419

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 #79

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,444
Percentile: 79.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #84. Votes: 2244. Organization: google. License: Apache 2.0.

79.3% percentile inside its fair comparison set

1,444Raw benchmark valueCI 1,432 - 1,456

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 #80

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,428
Percentile: 79%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: exclude_ties. Source rank: #87. Votes: 4155. Organization: google. License: Apache 2.0.

79% percentile inside its fair comparison set

1,428Raw benchmark valueCI 1,418 - 1,439

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 #90

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,438
Percentile: 76.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts. Source rank: #96. Votes: 3469. Organization: google. License: Apache 2.0.

76.3% percentile inside its fair comparison set

1,438Raw benchmark valueCI 1,428 - 1,448

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 #86

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,448
Percentile: 77.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts_english. Source rank: #90. Votes: 1272. Organization: google. License: Apache 2.0.

77.3% percentile inside its fair comparison set

1,448Raw benchmark valueCI 1,432 - 1,463

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 #80

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,420
Percentile: 79%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: instruction_following. Source rank: #85. Votes: 1737. Organization: google. License: Apache 2.0.

79% percentile inside its fair comparison set

1,420Raw benchmark valueCI 1,407 - 1,433

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 #88

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,427
Percentile: 75.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: longer_query. Source rank: #95. Votes: 1713. Organization: google. License: Apache 2.0.

75.5% percentile inside its fair comparison set

1,427Raw benchmark valueCI 1,414 - 1,441

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 #78

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,441
Percentile: 79.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: multi_turn. Source rank: #85. Votes: 1146. Organization: google. License: Apache 2.0.

79.4% percentile inside its fair comparison set

1,441Raw benchmark valueCI 1,424 - 1,458

Coding7 benchmarks33.8%

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 #138 · Source label: Gemma 4 26B A4B (Reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 13.6%
Percentile: 58.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `terminalbenchHard`.

58.7% percentile inside its fair comparison set

13.6%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 #142 · Source label: Gemma 4 26B A4B (Reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 40%
Percentile: 29.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `scicode`.

29.3% percentile inside its fair comparison set

40%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 #95

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,483
Percentile: 74.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: coding. Source rank: #104. Votes: 1443. Organization: google. License: Apache 2.0.

74.7% percentile inside its fair comparison set

1,483Raw benchmark valueCI 1,468 - 1,497

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 #100

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: coding. Source rank: #106. Votes: 1443. Organization: google. License: Apache 2.0.

73.3% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,432 - 1,461

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 #98

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,358
Percentile: 23.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #103. Votes: 1502. Organization: google. License: Apache 2.0.

23.6% percentile inside its fair comparison set

1,358Raw benchmark valueCI 1,342 - 1,375

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 #97

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,362
Percentile: 23.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: webdev-html. Source rank: #103. Votes: 205. Organization: google. License: Apache 2.0.

23.2% percentile inside its fair comparison set

1,362Raw benchmark valueCI 1,319 - 1,406

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 #92

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,350
Percentile: 19.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: webdev-react. Source rank: #99. Votes: 1291. Organization: google. License: Apache 2.0.

19.5% percentile inside its fair comparison set

1,350Raw benchmark valueCI 1,332 - 1,368

Reasoning / math / science5 benchmarks56%

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 #207 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 11.5%
Percentile: 61.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `hle`.

61.1% percentile inside its fair comparison set

11.5%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 #214 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 71.4%
Percentile: 56.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gpqa`.

56.8% percentile inside its fair comparison set

71.4%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 #350 · Source label: Gemma 4 26B A4B (Non-reasoning)

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

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 #51

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,468
Percentile: 86.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: math. Source rank: #53. Votes: 389. Organization: google. License: Apache 2.0.

86.1% percentile inside its fair comparison set

1,468Raw benchmark valueCI 1,440 - 1,496

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 #43

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,469
Percentile: 88.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: math. Source rank: #45. Votes: 389. Organization: google. License: Apache 2.0.

88.3% percentile inside its fair comparison set

1,469Raw benchmark valueCI 1,441 - 1,496

Professional reasoning18 benchmarksn/a

Text Arena · Expert

AR · Professional reasoning · Human

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

Rank #76

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,475
Percentile: 77.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: expert. Source rank: #81. Votes: 456. Organization: google. License: Apache 2.0.

77.1% percentile inside its fair comparison set

1,475Raw benchmark valueCI 1,450 - 1,501

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 #82

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,444
Percentile: 78%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_business_and_management_and_financial_operations. Source rank: #90. Votes: 1114. Organization: google. License: Apache 2.0.

78% percentile inside its fair comparison set

1,444Raw benchmark valueCI 1,427 - 1,461

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 #86

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,407
Percentile: 77.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_entertainment_and_sports_and_media. Source rank: #95. Votes: 1109. Organization: google. License: Apache 2.0.

77.3% percentile inside its fair comparison set

1,407Raw benchmark valueCI 1,390 - 1,425

Text Arena · Industry Legal And Government

AR · Professional reasoning · Human

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

Rank #60

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,464
Percentile: 83.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_legal_and_government. Source rank: #67. Votes: 416. Organization: google. License: Apache 2.0.

83.1% percentile inside its fair comparison set

1,464Raw benchmark valueCI 1,437 - 1,492

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 #92

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,455
Percentile: 75.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_life_and_physical_and_social_science. Source rank: #101. Votes: 925. Organization: google. License: Apache 2.0.

75.7% percentile inside its fair comparison set

1,455Raw benchmark valueCI 1,436 - 1,474

Text Arena · Industry Mathematical

AR · Professional reasoning · Human

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

Rank #45

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,478
Percentile: 87.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_mathematical. Source rank: #47. Votes: 309. Organization: google. License: Apache 2.0.

87.6% percentile inside its fair comparison set

1,478Raw benchmark valueCI 1,447 - 1,510

Text Arena · Industry Medicine And Healthcare

AR · Professional reasoning · Human

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

Rank #120

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,436
Percentile: 65.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_medicine_and_healthcare. Source rank: #134. Votes: 390. Organization: google. License: Apache 2.0.

65.6% percentile inside its fair comparison set

1,436Raw benchmark valueCI 1,407 - 1,465

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 #89

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,476
Percentile: 76.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_software_and_it_services. Source rank: #97. Votes: 2210. Organization: google. License: Apache 2.0.

76.6% percentile inside its fair comparison set

1,476Raw benchmark valueCI 1,464 - 1,488

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 #101

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,406
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_writing_and_literature_and_language. Source rank: #114. Votes: 1394. Organization: google. License: Apache 2.0.

73.3% percentile inside its fair comparison set

1,406Raw benchmark valueCI 1,391 - 1,421

Text Arena · Expert · No Style Control

AR · Professional reasoning · Human

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

Rank #81

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 `gemma-4-26b-a4b`. Category: expert. Source rank: #84. Votes: 456. Organization: google. License: Apache 2.0.

75.5% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,422 - 1,472

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 #78

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,429
Percentile: 79.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_business_and_management_and_financial_operations. Source rank: #81. Votes: 1114. Organization: google. License: Apache 2.0.

79.1% percentile inside its fair comparison set

1,429Raw benchmark valueCI 1,412 - 1,446

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 #80

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,403
Percentile: 78.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_entertainment_and_sports_and_media. Source rank: #87. Votes: 1109. Organization: google. License: Apache 2.0.

78.9% percentile inside its fair comparison set

1,403Raw benchmark valueCI 1,386 - 1,421

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 #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,457
Percentile: 85.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_legal_and_government. Source rank: #56. Votes: 416. Organization: google. License: Apache 2.0.

85.1% percentile inside its fair comparison set

1,457Raw benchmark valueCI 1,430 - 1,484

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 #83

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 78.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_life_and_physical_and_social_science. Source rank: #88. Votes: 925. Organization: google. License: Apache 2.0.

78.1% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,427 - 1,464

Text Arena · Industry Mathematical · No Style Control

AR · Professional reasoning · Human

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

Rank #42

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,475
Percentile: 88.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_mathematical. Source rank: #44. Votes: 309. Organization: google. License: Apache 2.0.

88.5% percentile inside its fair comparison set

1,475Raw benchmark valueCI 1,443 - 1,506

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 #112

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,427
Percentile: 67.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_medicine_and_healthcare. Source rank: #120. Votes: 390. Organization: google. License: Apache 2.0.

67.9% percentile inside its fair comparison set

1,427Raw benchmark valueCI 1,398 - 1,456

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 #88

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,454
Percentile: 76.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_software_and_it_services. Source rank: #92. Votes: 2210. Organization: google. License: Apache 2.0.

76.9% percentile inside its fair comparison set

1,454Raw benchmark valueCI 1,442 - 1,466

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 #97

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,405
Percentile: 74.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_writing_and_literature_and_language. Source rank: #105. Votes: 1394. Organization: google. License: Apache 2.0.

74.4% percentile inside its fair comparison set

1,405Raw benchmark valueCI 1,390 - 1,421

Search / tool use1 benchmark52.2%

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 #163 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 40.4%
Percentile: 52.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `tau2`.

52.2% percentile inside its fair comparison set

40.4%Raw benchmark value

Long context1 benchmark43.7%

Long Context Reasoning

AA · Long context · Objective

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

Rank #258 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 42.3%
Percentile: 43.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `lcr`.

43.7% percentile inside its fair comparison set

42.3%Raw benchmark value

Vision understanding19 benchmarks55.1%

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 #111 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 66.7%
Percentile: 47.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `mmmuPro`.

47.6% percentile inside its fair comparison set

66.7%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 #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,242
Percentile: 62.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #59. Votes: 23954. Organization: google. License: Apache 2.0.

62.6% percentile inside its fair comparison set

1,242Raw benchmark valueCI 1,235 - 1,248

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 #6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,242
Percentile: 83.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: captioning. Source rank: #6. Votes: 63. Organization: google. License: Apache 2.0.

83.9% percentile inside its fair comparison set

1,242Raw benchmark valueCI 1,170 - 1,313

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 #46

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,232
Percentile: 45.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing_vision. Source rank: #52. Votes: 1497. Organization: google. License: Apache 2.0.

45.1% percentile inside its fair comparison set

1,232Raw benchmark valueCI 1,216 - 1,248

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 #55

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,262
Percentile: 45.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: diagram. Source rank: #61. Votes: 6245. Organization: google. License: Apache 2.0.

45.5% percentile inside its fair comparison set

1,262Raw benchmark valueCI 1,253 - 1,270

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 #54

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,238
Percentile: 61.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #59. Votes: 9636. Organization: google. License: Apache 2.0.

61.6% percentile inside its fair comparison set

1,238Raw benchmark valueCI 1,230 - 1,247

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 #35

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,156
Percentile: 15%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: entity_recognition. Source rank: #44. Votes: 233. Organization: google. License: Apache 2.0.

15% percentile inside its fair comparison set

1,156Raw benchmark valueCI 1,120 - 1,193

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 #43

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 56.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: homework. Source rank: #47. Votes: 3840. Organization: google. License: Apache 2.0.

56.3% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,275 - 1,296

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 #35

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,245
Percentile: 53.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: humor. Source rank: #40. Votes: 997. Organization: google. License: Apache 2.0.

53.4% percentile inside its fair comparison set

1,245Raw benchmark valueCI 1,226 - 1,264

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 #51

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,258
Percentile: 50%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: ocr. Source rank: #57. Votes: 16888. Organization: google. License: Apache 2.0.

50% percentile inside its fair comparison set

1,258Raw benchmark valueCI 1,251 - 1,264

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 #49

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,260
Percentile: 65.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #53. Votes: 23954. Organization: google. License: Apache 2.0.

65.5% percentile inside its fair comparison set

1,260Raw benchmark valueCI 1,253 - 1,266

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 #4

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,300
Percentile: 90.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: captioning. Source rank: #4. Votes: 63. Organization: google. License: Apache 2.0.

90.3% percentile inside its fair comparison set

1,300Raw benchmark valueCI 1,230 - 1,369

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 #39

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,257
Percentile: 53.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing_vision. Source rank: #44. Votes: 1497. Organization: google. License: Apache 2.0.

53.7% percentile inside its fair comparison set

1,257Raw benchmark valueCI 1,241 - 1,273

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 #55

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,264
Percentile: 45.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: diagram. Source rank: #60. Votes: 6245. Organization: google. License: Apache 2.0.

45.5% percentile inside its fair comparison set

1,264Raw benchmark valueCI 1,255 - 1,273

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 #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,256
Percentile: 62.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #58. Votes: 9636. Organization: google. License: Apache 2.0.

62.3% percentile inside its fair comparison set

1,256Raw benchmark valueCI 1,247 - 1,264

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 #33

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,189
Percentile: 20%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: entity_recognition. Source rank: #41. Votes: 233. Organization: google. License: Apache 2.0.

20% percentile inside its fair comparison set

1,189Raw benchmark valueCI 1,153 - 1,226

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 #36

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,293
Percentile: 63.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: homework. Source rank: #40. Votes: 3840. Organization: google. License: Apache 2.0.

63.5% percentile inside its fair comparison set

1,293Raw benchmark valueCI 1,283 - 1,304

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 #27

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,272
Percentile: 64.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: humor. Source rank: #31. Votes: 997. Organization: google. License: Apache 2.0.

64.4% percentile inside its fair comparison set

1,272Raw benchmark valueCI 1,253 - 1,291

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 #48

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,268
Percentile: 53%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: ocr. Source rank: #52. Votes: 16888. Organization: google. License: Apache 2.0.

53% percentile inside its fair comparison set

1,268Raw benchmark valueCI 1,261 - 1,274

Multilingual12 benchmarksn/a

Text Arena · Chinese

AR · Multilingual · Human

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

Rank #70

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,493
Percentile: 80.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #74. Votes: 522. Organization: google. License: Apache 2.0.

80.4% percentile inside its fair comparison set

1,493Raw benchmark valueCI 1,469 - 1,518

Text Arena · French

AR · Multilingual · Human

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

Rank #64

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,470
Percentile: 75.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: french. Source rank: #72. Votes: 261. Organization: google. License: Apache 2.0.

75.5% percentile inside its fair comparison set

1,470Raw benchmark valueCI 1,433 - 1,508

Text Arena · Non English

AR · Multilingual · Human

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

Rank #92

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,423
Percentile: 75.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: non_english. Source rank: #104. Votes: 3837. Organization: google. License: Apache 2.0.

75.8% percentile inside its fair comparison set

1,423Raw benchmark valueCI 1,414 - 1,432

Text Arena · Russian

AR · Multilingual · Human

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

Rank #85

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 75.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: russian. Source rank: #94. Votes: 898. Organization: google. License: Apache 2.0.

75.4% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,419 - 1,456

Text Arena · Spanish

AR · Multilingual · Human

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

Rank #105

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,414
Percentile: 59.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: spanish. Source rank: #119. Votes: 202. Organization: google. License: Apache 2.0.

59.8% percentile inside its fair comparison set

1,414Raw benchmark valueCI 1,374 - 1,455

Text Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #55

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,495
Percentile: 84.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #58. Votes: 522. Organization: google. License: Apache 2.0.

84.7% percentile inside its fair comparison set

1,495Raw benchmark valueCI 1,470 - 1,520

Text Arena · French · No Style Control

AR · Multilingual · Human

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

Rank #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,459
Percentile: 79.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: french. Source rank: #57. Votes: 261. Organization: google. License: Apache 2.0.

79.8% percentile inside its fair comparison set

1,459Raw benchmark valueCI 1,422 - 1,496

Text Arena · Non English · No Style Control

AR · Multilingual · Human

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

Rank #79

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,421
Percentile: 79.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: non_english. Source rank: #86. Votes: 3837. Organization: google. License: Apache 2.0.

79.3% percentile inside its fair comparison set

1,421Raw benchmark valueCI 1,412 - 1,430

Text Arena · Russian · No Style Control

AR · Multilingual · Human

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

Rank #70

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,434
Percentile: 79.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: russian. Source rank: #76. Votes: 898. Organization: google. License: Apache 2.0.

79.8% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,416 - 1,452

Text Arena · Spanish · No Style Control

AR · Multilingual · Human

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

Rank #106

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,415
Percentile: 59.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: spanish. Source rank: #116. Votes: 202. Organization: google. License: Apache 2.0.

59.5% percentile inside its fair comparison set

1,415Raw benchmark valueCI 1,374 - 1,457

Vision Arena · Chinese

AR · Multilingual · Human

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

Rank #43

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,290
Percentile: 59.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #48. Votes: 1611. Organization: google. License: Apache 2.0.

59.6% percentile inside its fair comparison set

1,290Raw benchmark valueCI 1,273 - 1,307

Vision Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #38

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,313
Percentile: 64.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #42. Votes: 1611. Organization: google. License: Apache 2.0.

64.4% percentile inside its fair comparison set

1,313Raw benchmark valueCI 1,295 - 1,330

Source links and registry checks

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 →

Model profile · Google

Gemma 4 26B A4B

Open weightsmid · registry tag 2026 open multimodal

Thin verified coverage

Reads as thin verified coverage across the resolved source data. Capability fit excludes speed and cost.

Visible coverage: 13.7%
Verified coverage: 13.7%
Spread: 58.2%
Last verified: Oct 3, 2026

52%capability fit

textvisioncodedocument5 aliases39 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

Thin verified coverageThis model currently reads as thin verified coverage across the resolved source data.

Chat / text27 benchmarks56.4%

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 #240 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 13
Percentile: 56.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `intelligenceIndex`.

56.9% percentile inside its fair comparison set

13Raw 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 #317 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 15.5%
Percentile: 30.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceAccuracy`.

30.2% percentile inside its fair comparison set

15.5%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 #373 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 7.9%
Percentile: 17.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceNonHallucination`.

17.9% percentile inside its fair comparison set

7.9%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 #137 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 45.4%
Percentile: 60.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `ifbench`.

60.9% percentile inside its fair comparison set

45.4%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 #70 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $0.1 /1M input tokens
Percentile: 80.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mInputTokens`.

80.5% percentile inside its fair comparison set

$0.1 /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 #68 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $0.4 /1M output tokens
Percentile: 81.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mOutputTokens`.

81.4% percentile inside its fair comparison set

$0.4 /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 #95 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 109.1 tokens/s
Percentile: 64.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianOutputTokensPerSecond`.

64.1% percentile inside its fair comparison set

109.1 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 #49 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 1.23s
Percentile: 81.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstTokenSeconds`.

81.8% percentile inside its fair comparison set

1.23sRaw 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 #29 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 1.23s
Percentile: 89.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstAnswerTokenSeconds`.

89.3% percentile inside its fair comparison set

1.23sRaw 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 #91

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 76.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #100. Votes: 6081. Organization: google. License: Apache 2.0.

76.1% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,430 - 1,445

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 #101

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,400
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing. Source rank: #111. Votes: 995. Organization: google. License: Apache 2.0.

73.3% percentile inside its fair comparison set

1,400Raw benchmark valueCI 1,381 - 1,418

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 #86

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,450
Percentile: 77.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #94. Votes: 2244. Organization: google. License: Apache 2.0.

77.4% percentile inside its fair comparison set

1,450Raw benchmark valueCI 1,438 - 1,462

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 #90

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,434
Percentile: 76.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: exclude_ties. Source rank: #99. Votes: 4155. Organization: google. License: Apache 2.0.

76.3% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,423 - 1,445

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 #88

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,461
Percentile: 76.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts. Source rank: #96. Votes: 3469. Organization: google. License: Apache 2.0.

76.9% percentile inside its fair comparison set

1,461Raw benchmark valueCI 1,451 - 1,470

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 #77

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,471
Percentile: 79.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts_english. Source rank: #84. Votes: 1272. Organization: google. License: Apache 2.0.

79.7% percentile inside its fair comparison set

1,471Raw benchmark valueCI 1,455 - 1,487

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 #76

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 80.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: instruction_following. Source rank: #83. Votes: 1737. Organization: google. License: Apache 2.0.

80.1% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,424 - 1,451

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 #89

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 75.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: longer_query. Source rank: #98. Votes: 1713. Organization: google. License: Apache 2.0.

75.2% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,432 - 1,460

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 #84

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,448
Percentile: 77.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: multi_turn. Source rank: #94. Votes: 1146. Organization: google. License: Apache 2.0.

77.8% percentile inside its fair comparison set

1,448Raw benchmark valueCI 1,431 - 1,465

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 #82

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 `gemma-4-26b-a4b`. Category: overall. Source rank: #89. Votes: 6081. Organization: google. License: Apache 2.0.

78.5% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,426 - 1,441

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 #89

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,401
Percentile: 76.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing. Source rank: #98. Votes: 995. Organization: google. License: Apache 2.0.

76.5% percentile inside its fair comparison set

1,401Raw benchmark valueCI 1,383 - 1,419

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 #79

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,444
Percentile: 79.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #84. Votes: 2244. Organization: google. License: Apache 2.0.

79.3% percentile inside its fair comparison set

1,444Raw benchmark valueCI 1,432 - 1,456

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 #80

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,428
Percentile: 79%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: exclude_ties. Source rank: #87. Votes: 4155. Organization: google. License: Apache 2.0.

79% percentile inside its fair comparison set

1,428Raw benchmark valueCI 1,418 - 1,439

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 #90

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,438
Percentile: 76.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts. Source rank: #96. Votes: 3469. Organization: google. License: Apache 2.0.

76.3% percentile inside its fair comparison set

1,438Raw benchmark valueCI 1,428 - 1,448

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 #86

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,448
Percentile: 77.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: hard_prompts_english. Source rank: #90. Votes: 1272. Organization: google. License: Apache 2.0.

77.3% percentile inside its fair comparison set

1,448Raw benchmark valueCI 1,432 - 1,463

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 #80

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,420
Percentile: 79%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: instruction_following. Source rank: #85. Votes: 1737. Organization: google. License: Apache 2.0.

79% percentile inside its fair comparison set

1,420Raw benchmark valueCI 1,407 - 1,433

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 #88

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,427
Percentile: 75.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: longer_query. Source rank: #95. Votes: 1713. Organization: google. License: Apache 2.0.

75.5% percentile inside its fair comparison set

1,427Raw benchmark valueCI 1,414 - 1,441

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 #78

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,441
Percentile: 79.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: multi_turn. Source rank: #85. Votes: 1146. Organization: google. License: Apache 2.0.

79.4% percentile inside its fair comparison set

1,441Raw benchmark valueCI 1,424 - 1,458

Coding7 benchmarks33.8%

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 #138 · Source label: Gemma 4 26B A4B (Reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 13.6%
Percentile: 58.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `terminalbenchHard`.

58.7% percentile inside its fair comparison set

13.6%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 #142 · Source label: Gemma 4 26B A4B (Reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 40%
Percentile: 29.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `scicode`.

29.3% percentile inside its fair comparison set

40%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 #95

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,483
Percentile: 74.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: coding. Source rank: #104. Votes: 1443. Organization: google. License: Apache 2.0.

74.7% percentile inside its fair comparison set

1,483Raw benchmark valueCI 1,468 - 1,497

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 #100

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: coding. Source rank: #106. Votes: 1443. Organization: google. License: Apache 2.0.

73.3% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,432 - 1,461

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 #98

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,358
Percentile: 23.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #103. Votes: 1502. Organization: google. License: Apache 2.0.

23.6% percentile inside its fair comparison set

1,358Raw benchmark valueCI 1,342 - 1,375

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 #97

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,362
Percentile: 23.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: webdev-html. Source rank: #103. Votes: 205. Organization: google. License: Apache 2.0.

23.2% percentile inside its fair comparison set

1,362Raw benchmark valueCI 1,319 - 1,406

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 #92

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,350
Percentile: 19.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: webdev-react. Source rank: #99. Votes: 1291. Organization: google. License: Apache 2.0.

19.5% percentile inside its fair comparison set

1,350Raw benchmark valueCI 1,332 - 1,368

Reasoning / math / science5 benchmarks56%

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 #207 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 11.5%
Percentile: 61.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `hle`.

61.1% percentile inside its fair comparison set

11.5%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 #214 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 71.4%
Percentile: 56.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gpqa`.

56.8% percentile inside its fair comparison set

71.4%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 #350 · Source label: Gemma 4 26B A4B (Non-reasoning)

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

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 #51

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,468
Percentile: 86.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: math. Source rank: #53. Votes: 389. Organization: google. License: Apache 2.0.

86.1% percentile inside its fair comparison set

1,468Raw benchmark valueCI 1,440 - 1,496

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 #43

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,469
Percentile: 88.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: math. Source rank: #45. Votes: 389. Organization: google. License: Apache 2.0.

88.3% percentile inside its fair comparison set

1,469Raw benchmark valueCI 1,441 - 1,496

Professional reasoning18 benchmarksn/a

Text Arena · Expert

AR · Professional reasoning · Human

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

Rank #76

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,475
Percentile: 77.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: expert. Source rank: #81. Votes: 456. Organization: google. License: Apache 2.0.

77.1% percentile inside its fair comparison set

1,475Raw benchmark valueCI 1,450 - 1,501

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 #82

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,444
Percentile: 78%
Last updated: recent
Eligibility: headline eligible

78% percentile inside its fair comparison set

1,444Raw benchmark valueCI 1,427 - 1,461

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 #86

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,407
Percentile: 77.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_entertainment_and_sports_and_media. Source rank: #95. Votes: 1109. Organization: google. License: Apache 2.0.

77.3% percentile inside its fair comparison set

1,407Raw benchmark valueCI 1,390 - 1,425

Text Arena · Industry Legal And Government

AR · Professional reasoning · Human

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

Rank #60

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,464
Percentile: 83.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_legal_and_government. Source rank: #67. Votes: 416. Organization: google. License: Apache 2.0.

83.1% percentile inside its fair comparison set

1,464Raw benchmark valueCI 1,437 - 1,492

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 #92

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,455
Percentile: 75.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_life_and_physical_and_social_science. Source rank: #101. Votes: 925. Organization: google. License: Apache 2.0.

75.7% percentile inside its fair comparison set

1,455Raw benchmark valueCI 1,436 - 1,474

Text Arena · Industry Mathematical

AR · Professional reasoning · Human

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

Rank #45

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,478
Percentile: 87.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_mathematical. Source rank: #47. Votes: 309. Organization: google. License: Apache 2.0.

87.6% percentile inside its fair comparison set

1,478Raw benchmark valueCI 1,447 - 1,510

Text Arena · Industry Medicine And Healthcare

AR · Professional reasoning · Human

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

Rank #120

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,436
Percentile: 65.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_medicine_and_healthcare. Source rank: #134. Votes: 390. Organization: google. License: Apache 2.0.

65.6% percentile inside its fair comparison set

1,436Raw benchmark valueCI 1,407 - 1,465

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 #89

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,476
Percentile: 76.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_software_and_it_services. Source rank: #97. Votes: 2210. Organization: google. License: Apache 2.0.

76.6% percentile inside its fair comparison set

1,476Raw benchmark valueCI 1,464 - 1,488

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 #101

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,406
Percentile: 73.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_writing_and_literature_and_language. Source rank: #114. Votes: 1394. Organization: google. License: Apache 2.0.

73.3% percentile inside its fair comparison set

1,406Raw benchmark valueCI 1,391 - 1,421

Text Arena · Expert · No Style Control

AR · Professional reasoning · Human

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

Rank #81

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 `gemma-4-26b-a4b`. Category: expert. Source rank: #84. Votes: 456. Organization: google. License: Apache 2.0.

75.5% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,422 - 1,472

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 #78

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,429
Percentile: 79.1%
Last updated: recent
Eligibility: headline eligible

79.1% percentile inside its fair comparison set

1,429Raw benchmark valueCI 1,412 - 1,446

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 #80

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,403
Percentile: 78.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_entertainment_and_sports_and_media. Source rank: #87. Votes: 1109. Organization: google. License: Apache 2.0.

78.9% percentile inside its fair comparison set

1,403Raw benchmark valueCI 1,386 - 1,421

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 #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,457
Percentile: 85.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_legal_and_government. Source rank: #56. Votes: 416. Organization: google. License: Apache 2.0.

85.1% percentile inside its fair comparison set

1,457Raw benchmark valueCI 1,430 - 1,484

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 #83

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 78.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_life_and_physical_and_social_science. Source rank: #88. Votes: 925. Organization: google. License: Apache 2.0.

78.1% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,427 - 1,464

Text Arena · Industry Mathematical · No Style Control

AR · Professional reasoning · Human

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

Rank #42

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,475
Percentile: 88.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_mathematical. Source rank: #44. Votes: 309. Organization: google. License: Apache 2.0.

88.5% percentile inside its fair comparison set

1,475Raw benchmark valueCI 1,443 - 1,506

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 #112

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,427
Percentile: 67.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_medicine_and_healthcare. Source rank: #120. Votes: 390. Organization: google. License: Apache 2.0.

67.9% percentile inside its fair comparison set

1,427Raw benchmark valueCI 1,398 - 1,456

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 #88

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,454
Percentile: 76.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_software_and_it_services. Source rank: #92. Votes: 2210. Organization: google. License: Apache 2.0.

76.9% percentile inside its fair comparison set

1,454Raw benchmark valueCI 1,442 - 1,466

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 #97

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,405
Percentile: 74.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: industry_writing_and_literature_and_language. Source rank: #105. Votes: 1394. Organization: google. License: Apache 2.0.

74.4% percentile inside its fair comparison set

1,405Raw benchmark valueCI 1,390 - 1,421

Search / tool use1 benchmark52.2%

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 #163 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 40.4%
Percentile: 52.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `tau2`.

52.2% percentile inside its fair comparison set

40.4%Raw benchmark value

Long context1 benchmark43.7%

Long Context Reasoning

AA · Long context · Objective

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

Rank #258 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 42.3%
Percentile: 43.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `lcr`.

43.7% percentile inside its fair comparison set

42.3%Raw benchmark value

Vision understanding19 benchmarks55.1%

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 #111 · Source label: Gemma 4 26B A4B (Non-reasoning)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 66.7%
Percentile: 47.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `mmmuPro`.

47.6% percentile inside its fair comparison set

66.7%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 #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,242
Percentile: 62.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #59. Votes: 23954. Organization: google. License: Apache 2.0.

62.6% percentile inside its fair comparison set

1,242Raw benchmark valueCI 1,235 - 1,248

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 #6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,242
Percentile: 83.9%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: captioning. Source rank: #6. Votes: 63. Organization: google. License: Apache 2.0.

83.9% percentile inside its fair comparison set

1,242Raw benchmark valueCI 1,170 - 1,313

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 #46

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,232
Percentile: 45.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing_vision. Source rank: #52. Votes: 1497. Organization: google. License: Apache 2.0.

45.1% percentile inside its fair comparison set

1,232Raw benchmark valueCI 1,216 - 1,248

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 #55

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,262
Percentile: 45.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: diagram. Source rank: #61. Votes: 6245. Organization: google. License: Apache 2.0.

45.5% percentile inside its fair comparison set

1,262Raw benchmark valueCI 1,253 - 1,270

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 #54

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,238
Percentile: 61.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #59. Votes: 9636. Organization: google. License: Apache 2.0.

61.6% percentile inside its fair comparison set

1,238Raw benchmark valueCI 1,230 - 1,247

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 #35

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,156
Percentile: 15%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: entity_recognition. Source rank: #44. Votes: 233. Organization: google. License: Apache 2.0.

15% percentile inside its fair comparison set

1,156Raw benchmark valueCI 1,120 - 1,193

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 #43

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 56.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: homework. Source rank: #47. Votes: 3840. Organization: google. License: Apache 2.0.

56.3% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,275 - 1,296

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 #35

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,245
Percentile: 53.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: humor. Source rank: #40. Votes: 997. Organization: google. License: Apache 2.0.

53.4% percentile inside its fair comparison set

1,245Raw benchmark valueCI 1,226 - 1,264

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 #51

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,258
Percentile: 50%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: ocr. Source rank: #57. Votes: 16888. Organization: google. License: Apache 2.0.

50% percentile inside its fair comparison set

1,258Raw benchmark valueCI 1,251 - 1,264

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 #49

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,260
Percentile: 65.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: overall. Source rank: #53. Votes: 23954. Organization: google. License: Apache 2.0.

65.5% percentile inside its fair comparison set

1,260Raw benchmark valueCI 1,253 - 1,266

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 #4

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,300
Percentile: 90.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: captioning. Source rank: #4. Votes: 63. Organization: google. License: Apache 2.0.

90.3% percentile inside its fair comparison set

1,300Raw benchmark valueCI 1,230 - 1,369

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 #39

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,257
Percentile: 53.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: creative_writing_vision. Source rank: #44. Votes: 1497. Organization: google. License: Apache 2.0.

53.7% percentile inside its fair comparison set

1,257Raw benchmark valueCI 1,241 - 1,273

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 #55

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,264
Percentile: 45.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: diagram. Source rank: #60. Votes: 6245. Organization: google. License: Apache 2.0.

45.5% percentile inside its fair comparison set

1,264Raw benchmark valueCI 1,255 - 1,273

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 #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,256
Percentile: 62.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: english. Source rank: #58. Votes: 9636. Organization: google. License: Apache 2.0.

62.3% percentile inside its fair comparison set

1,256Raw benchmark valueCI 1,247 - 1,264

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 #33

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,189
Percentile: 20%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: entity_recognition. Source rank: #41. Votes: 233. Organization: google. License: Apache 2.0.

20% percentile inside its fair comparison set

1,189Raw benchmark valueCI 1,153 - 1,226

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 #36

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,293
Percentile: 63.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: homework. Source rank: #40. Votes: 3840. Organization: google. License: Apache 2.0.

63.5% percentile inside its fair comparison set

1,293Raw benchmark valueCI 1,283 - 1,304

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 #27

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,272
Percentile: 64.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: humor. Source rank: #31. Votes: 997. Organization: google. License: Apache 2.0.

64.4% percentile inside its fair comparison set

1,272Raw benchmark valueCI 1,253 - 1,291

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 #48

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,268
Percentile: 53%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: ocr. Source rank: #52. Votes: 16888. Organization: google. License: Apache 2.0.

53% percentile inside its fair comparison set

1,268Raw benchmark valueCI 1,261 - 1,274

Multilingual12 benchmarksn/a

Text Arena · Chinese

AR · Multilingual · Human

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

Rank #70

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,493
Percentile: 80.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #74. Votes: 522. Organization: google. License: Apache 2.0.

80.4% percentile inside its fair comparison set

1,493Raw benchmark valueCI 1,469 - 1,518

Text Arena · French

AR · Multilingual · Human

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

Rank #64

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,470
Percentile: 75.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: french. Source rank: #72. Votes: 261. Organization: google. License: Apache 2.0.

75.5% percentile inside its fair comparison set

1,470Raw benchmark valueCI 1,433 - 1,508

Text Arena · Non English

AR · Multilingual · Human

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

Rank #92

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,423
Percentile: 75.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: non_english. Source rank: #104. Votes: 3837. Organization: google. License: Apache 2.0.

75.8% percentile inside its fair comparison set

1,423Raw benchmark valueCI 1,414 - 1,432

Text Arena · Russian

AR · Multilingual · Human

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

Rank #85

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 75.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: russian. Source rank: #94. Votes: 898. Organization: google. License: Apache 2.0.

75.4% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,419 - 1,456

Text Arena · Spanish

AR · Multilingual · Human

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

Rank #105

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,414
Percentile: 59.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: spanish. Source rank: #119. Votes: 202. Organization: google. License: Apache 2.0.

59.8% percentile inside its fair comparison set

1,414Raw benchmark valueCI 1,374 - 1,455

Text Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #55

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,495
Percentile: 84.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #58. Votes: 522. Organization: google. License: Apache 2.0.

84.7% percentile inside its fair comparison set

1,495Raw benchmark valueCI 1,470 - 1,520

Text Arena · French · No Style Control

AR · Multilingual · Human

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

Rank #53

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,459
Percentile: 79.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: french. Source rank: #57. Votes: 261. Organization: google. License: Apache 2.0.

79.8% percentile inside its fair comparison set

1,459Raw benchmark valueCI 1,422 - 1,496

Text Arena · Non English · No Style Control

AR · Multilingual · Human

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

Rank #79

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,421
Percentile: 79.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: non_english. Source rank: #86. Votes: 3837. Organization: google. License: Apache 2.0.

79.3% percentile inside its fair comparison set

1,421Raw benchmark valueCI 1,412 - 1,430

Text Arena · Russian · No Style Control

AR · Multilingual · Human

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

Rank #70

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,434
Percentile: 79.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: russian. Source rank: #76. Votes: 898. Organization: google. License: Apache 2.0.

79.8% percentile inside its fair comparison set

1,434Raw benchmark valueCI 1,416 - 1,452

Text Arena · Spanish · No Style Control

AR · Multilingual · Human

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

Rank #106

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,415
Percentile: 59.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: spanish. Source rank: #116. Votes: 202. Organization: google. License: Apache 2.0.

59.5% percentile inside its fair comparison set

1,415Raw benchmark valueCI 1,374 - 1,457

Vision Arena · Chinese

AR · Multilingual · Human

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

Rank #43

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,290
Percentile: 59.6%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #48. Votes: 1611. Organization: google. License: Apache 2.0.

59.6% percentile inside its fair comparison set

1,290Raw benchmark valueCI 1,273 - 1,307

Vision Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #38

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,313
Percentile: 64.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `gemma-4-26b-a4b`. Category: chinese. Source rank: #42. Votes: 1611. Organization: google. License: Apache 2.0.

64.4% percentile inside its fair comparison set

1,313Raw benchmark valueCI 1,295 - 1,330

Source links and registry checks

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 →

Gemma 4 26B A4B

Current snapshot.

Source-linked scores by benchmark

Source links and registry checks

Loading model evidence.

Gemma 4 26B A4B

Current snapshot.

Source-linked scores by benchmark

Source links and registry checks