Model profile · Anthropic

Claude Sonnet 4.6

Closed weightspremium · registry tag 2026 workhorse

Thin verified coverage

Reads as thin verified coverage across the resolved source data.

Visible coverage: 52.1%
Verified coverage: 47.3%
Spread: 97.8%
Last verified: Aug 19, 2026

66%bench fit

textcodevisiondocumentsearch10 aliases47 official source links

Open compare

Data version

Current snapshot.

Published Aug 19, 2026Model list checked2 stale sources216 claim warnings9 providers · 1149 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.

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

Chat / text28 benchmarks74.4%

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 #66 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 35
Percentile: 86.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `intelligenceIndex`.

86.1% percentile inside its fair comparison set

35Raw 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 #42 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 36.5%
Percentile: 88.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceAccuracy`.

88.8% percentile inside its fair comparison set

36.5%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 #99 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 31.5%
Percentile: 73.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceNonHallucination`.

73.2% percentile inside its fair comparison set

31.5%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 #179 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 41.2%
Percentile: 48.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `ifbench`.

48.4% percentile inside its fair comparison set

41.2%Raw benchmark value

Blended 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 #266 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $6 /1M tokens
Percentile: 12.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mBlended0To3To1`.

12.3% percentile inside its fair comparison set

$6 /1M tokensRaw 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 #266 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $3 /1M input tokens
Percentile: 12.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mInputTokens`.

12.3% percentile inside its fair comparison set

$3 /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 #267 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $15 /1M output tokens
Percentile: 12.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mOutputTokens`.

12.3% percentile inside its fair comparison set

$15 /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 #197 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 43.2 tokens/s
Percentile: 12.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianOutputTokensPerSecond`.

12.5% percentile inside its fair comparison set

43.2 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 #220 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 149.24s
Percentile: 2.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstTokenSeconds`.

2.2% percentile inside its fair comparison set

149.24sRaw 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 #220 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 149.24s
Percentile: 2.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstAnswerTokenSeconds`.

2.2% percentile inside its fair comparison set

149.24sRaw 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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,472
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #24. Votes: 39561. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,472Raw benchmark valueCI 1,468 - 1,476

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 #20

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 94.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing. Source rank: #28. Votes: 6301. Organization: anthropic. License: Proprietary.

94.1% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,438 - 1,455

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,485
Percentile: 97.2%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #12. Votes: 18942. Organization: anthropic. License: Proprietary.

97.2% percentile inside its fair comparison set

1,485Raw benchmark valueCI 1,479 - 1,490

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,481
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: exclude_ties. Source rank: #23. Votes: 29825. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,481Raw benchmark valueCI 1,476 - 1,487

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,504
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts. Source rank: #10. Votes: 25042. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,504Raw benchmark valueCI 1,499 - 1,509

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 #6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,509
Percentile: 98.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts_english. Source rank: #8. Votes: 12541. Organization: anthropic. License: Proprietary.

98.5% percentile inside its fair comparison set

1,509Raw benchmark valueCI 1,503 - 1,516

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,479
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: instruction_following. Source rank: #10. Votes: 12635. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,479Raw benchmark valueCI 1,473 - 1,485

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,495
Percentile: 97.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: longer_query. Source rank: #11. Votes: 15661. Organization: anthropic. License: Proprietary.

97.7% percentile inside its fair comparison set

1,495Raw benchmark valueCI 1,488 - 1,501

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 #14

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,484
Percentile: 96%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: multi_turn. Source rank: #16. Votes: 7169. Organization: anthropic. License: Proprietary.

96% percentile inside its fair comparison set

1,484Raw benchmark valueCI 1,477 - 1,492

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #22. Votes: 39561. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,452 - 1,461

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 #26

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,432
Percentile: 92.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing. Source rank: #34. Votes: 6301. Organization: anthropic. License: Proprietary.

92.3% percentile inside its fair comparison set

1,432Raw benchmark valueCI 1,424 - 1,440

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 #13

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,472
Percentile: 96.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #15. Votes: 18942. Organization: anthropic. License: Proprietary.

96.3% percentile inside its fair comparison set

1,472Raw benchmark valueCI 1,467 - 1,478

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: exclude_ties. Source rank: #22. Votes: 29825. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,451 - 1,462

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,484
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts. Source rank: #10. Votes: 25042. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,484Raw benchmark valueCI 1,479 - 1,489

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,494
Percentile: 98.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts_english. Source rank: #7. Votes: 12541. Organization: anthropic. License: Proprietary.

98.8% percentile inside its fair comparison set

1,494Raw benchmark valueCI 1,488 - 1,501

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,479
Percentile: 98.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: instruction_following. Source rank: #7. Votes: 12635. Organization: anthropic. License: Proprietary.

98.8% percentile inside its fair comparison set

1,479Raw benchmark valueCI 1,472 - 1,485

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,481
Percentile: 97%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: longer_query. Source rank: #13. Votes: 15661. Organization: anthropic. License: Proprietary.

97% percentile inside its fair comparison set

1,481Raw benchmark valueCI 1,475 - 1,487

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,466
Percentile: 94.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: multi_turn. Source rank: #22. Votes: 7169. Organization: anthropic. License: Proprietary.

94.7% percentile inside its fair comparison set

1,466Raw benchmark valueCI 1,458 - 1,474

Coding30 benchmarks77%

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 #19 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 42.4%
Percentile: 94.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `terminalbenchHard`.

94.5% percentile inside its fair comparison set

42.4%Raw benchmark value

SciCode

AA · Coding · Objective

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

Rank #52 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 44.1%
Percentile: 88.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `scicode`.

88.5% percentile inside its fair comparison set

44.1%Raw benchmark value

Coding Index

AA · Coding · Combined

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

Rank #31 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 63
Percentile: 84%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `codingIndex`.

84% percentile inside its fair comparison set

63Raw benchmark value

Agentic Index

AA · Coding · Combined

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

Rank #21 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 42
Percentile: 85%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `agenticIndex`.

85% percentile inside its fair comparison set

42Raw benchmark value

Code 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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,521
Percentile: 90.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #12. Votes: 13612. Organization: anthropic. License: Proprietary.

90.4% percentile inside its fair comparison set

1,521Raw benchmark valueCI 1,515 - 1,527

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,521
Percentile: 90.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: webdev. Source rank: #12. Votes: 13612. Organization: anthropic. License: Proprietary.

90.4% percentile inside its fair comparison set

1,521Raw benchmark valueCI 1,515 - 1,527

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,515
Percentile: 89%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: webdev-html. Source rank: #12. Votes: 1879. Organization: anthropic. License: Proprietary.

89% percentile inside its fair comparison set

1,515Raw benchmark valueCI 1,501 - 1,529

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,515
Percentile: 88.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: webdev-react. Source rank: #12. Votes: 11688. Organization: anthropic. License: Proprietary.

88.1% percentile inside its fair comparison set

1,515Raw benchmark valueCI 1,509 - 1,522

Code Migration

VALS-AI · Coding · Objective

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

Rank #5 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 39.9%
Percentile: 81.8%
Last updated: aging
Eligibility: headline eligible

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

81.8% percentile inside its fair comparison set

39.9%Raw benchmark valueCI 31.8% - 48%

LiveCodeBench

VALS-AI · Coding · Objective

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

Rank #34 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 82.1%
Percentile: 63.3%
Last updated: aging
Eligibility: headline eligible

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

63.3% percentile inside its fair comparison set

82.1%Raw benchmark valueCI 80% - 84.2%

ProgramBench

VALS-AI · Coding · Objective

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

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 0.5%
Percentile: 90%
Last updated: aging
Eligibility: headline eligible

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

90% percentile inside its fair comparison set

0.5%Raw benchmark valueCI 0% - 1.5%

SWE-bench Verified

VALS-AI · Coding · Objective

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

Rank #12 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 77.4%
Percentile: 79.6%
Last updated: aging
Eligibility: headline eligible

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

79.6% percentile inside its fair comparison set

77.4%Raw benchmark valueCI 73.7% - 81.1%

Terminal-Bench 2.1

VALS-AI · Coding · Objective

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

Rank #11 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 57.3%
Percentile: 63%
Last updated: aging
Eligibility: headline eligible

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

63% percentile inside its fair comparison set

57.3%Raw benchmark valueCI 56% - 58.6%

Vibe Code Bench v1.1

VALS-AI · Coding · Objective

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

Rank #9 · Source label: anthropic/claude-sonnet-4-6-claude-code

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 55.8%
Percentile: 83.7%
Last updated: aging
Eligibility: headline eligible

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

83.7% percentile inside its fair comparison set

55.8%Raw benchmark valueCI 46.6% - 65%

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,527
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: coding. Source rank: #10. Votes: 10802. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,527Raw benchmark valueCI 1,520 - 1,534

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,502
Percentile: 98.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: coding. Source rank: #7. Votes: 10802. Organization: anthropic. License: Proprietary.

98.8% percentile inside its fair comparison set

1,502Raw benchmark valueCI 1,495 - 1,508

IOI

VALS-AI · Coding · Objective

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

Rank #31 · Source label: anthropic/claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 6.5%
Percentile: 34.1%
Last updated: aging
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: ioi; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

34.1% percentile inside its fair comparison set

6.5%Raw benchmark valueCI 2.4% - 10.6%

Code Arena · Image To Webdev

AR · Coding · Human

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

Rank #2

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,557
Percentile: 94.1%
Last updated: stale
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: image_to_webdev. Source rank: #2. Votes: 3158. Organization: anthropic. License: Proprietary.

94.1% percentile inside its fair comparison set

1,557Raw benchmark valueCI 1,544 - 1,570

Debugging

BB · Coding · Rubric

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

Rank #3 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 86.6%
Percentile: 91.7%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-debugging.

91.7% percentile inside its fair comparison set

86.6%Raw benchmark value

Refactoring

BB · Coding · Rubric

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

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 69.4%
Percentile: 76.2%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-refactoring.

76.2% percentile inside its fair comparison set

69.4%Raw benchmark value

Security

BB · Coding · Rubric

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

Rank #2 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 85.3%
Percentile: 100%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-security.

100% percentile inside its fair comparison set

85.3%Raw benchmark value

UI bench

BB · Coding · Rubric

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

Rank #1 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 81.5%
Percentile: 100%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ui.

100% percentile inside its fair comparison set

81.5%Raw benchmark value

Speed throughput

BB · Coding · Speed / cost

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

Rank #13 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 95.3 t/s
Percentile: 52%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-throughput.

52% percentile inside its fair comparison set

95.3 t/sRaw benchmark value

Speed TTFT

BB · Coding · Speed / cost

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

Rank #9 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 1207.00ms
Percentile: 68%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ttft.

68% percentile inside its fair comparison set

1207.00msRaw benchmark value

Speed cost

BB · Coding · Speed / cost

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

Rank #22 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: $0.5 $
Percentile: 16%
Last updated: stale
Eligibility: headline eligible

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

16% percentile inside its fair comparison set

$0.5 $Raw benchmark value

Terminal-Bench 2.0

OFF · Coding · Objective

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

Rank #6

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 59.1%
Percentile: 16.7%
Last updated: stale
Eligibility: headline eligible

Verified against Anthropic's public Claude Sonnet 4.6 system card, checked on April 29, 2026. Uses the published Terminal-Bench 2.0 figure. Checked 2026-04-29. Verification: manual_public_page_verification.

16.7% percentile inside its fair comparison set

59.1%Raw benchmark value

SWE-Bench Verified

OFF · Coding · Objective

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

Rank #4

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 79.6%
Percentile: 40%
Last updated: stale
Eligibility: headline eligible

Verified against Anthropic's public Claude Sonnet 4.6 system card, checked on April 29, 2026. Uses the published SWE-Bench Verified figure. Checked 2026-04-29. Verification: manual_public_page_verification.

40% percentile inside its fair comparison set

79.6%Raw benchmark value

Overall quality

BB · Coding · Combined

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

Rank #5 · Source label: Claude Sonnet 4.6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 89.5%
Percentile: 81.8%
Last updated: archived
Eligibility: headline eligible

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

81.8% percentile inside its fair comparison set

89.5%Raw benchmark value

Overall vibe

BB · Coding · Combined

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

Rank #7 · Source label: Claude Sonnet 4.6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 82.7%
Percentile: 72.7%
Last updated: archived
Eligibility: headline eligible

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

72.7% percentile inside its fair comparison set

82.7%Raw benchmark value

Overall coverage

BB · Coding · Combined

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

Rank #6 · Source label: Claude Sonnet 4.6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

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

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

100% percentile inside its fair comparison set

100%Raw benchmark value

Reasoning / math / science11 benchmarks65.6%

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 #133 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 11.2%
Percentile: 70.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `hle`.

70.1% percentile inside its fair comparison set

11.2%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 #100 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 79.7%
Percentile: 78%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gpqa`.

78% percentile inside its fair comparison set

79.7%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 #86 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 0.9%
Percentile: 77%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `critpt`.

77% percentile inside its fair comparison set

0.9%Raw benchmark value

ProofBench

VALS-AI · Reasoning / math / science · Objective

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

Rank #7 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 45%
Percentile: 82.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Anthropic.

82.9% percentile inside its fair comparison set

45%Raw benchmark valueCI 35.2% - 54.8%

GPQA Diamond

VALS-AI · Reasoning / math / science · Objective

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

Rank #28 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 85.6%
Percentile: 71.9%
Last updated: aging
Eligibility: headline eligible

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

71.9% percentile inside its fair comparison set

85.6%Raw benchmark valueCI 80.9% - 90.3%

MMLU Pro

VALS-AI · Reasoning / math / science · Objective

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

Rank #16 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 87.3%
Percentile: 83.1%
Last updated: aging
Eligibility: headline eligible

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

83.1% percentile inside its fair comparison set

87.3%Raw benchmark valueCI 86.5% - 88.2%

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 #27

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,459
Percentile: 91.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: math. Source rank: #33. Votes: 2296. Organization: anthropic. License: Proprietary.

91.7% percentile inside its fair comparison set

1,459Raw benchmark valueCI 1,446 - 1,471

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 #25

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 92.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: math. Source rank: #31. Votes: 2296. Organization: anthropic. License: Proprietary.

92.4% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,443 - 1,469

Reasoning

BB · Reasoning / math / science · Rubric

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

Rank #15 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 37.2%
Percentile: 33.3%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-reasoning.

33.3% percentile inside its fair comparison set

37.2%Raw benchmark value

Humanity's Last Exam

OFF · Reasoning / math / science · Objective

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

Rank #7

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 33.2%
Percentile: 14.3%
Last updated: stale
Eligibility: headline eligible

Verified against Anthropic's public Claude Sonnet 4.6 system card, checked on April 29, 2026. Uses the published no-tools Humanity's Last Exam figure. Checked 2026-04-29. Verification: manual_public_page_verification.

14.3% percentile inside its fair comparison set

33.2%Raw benchmark value

EnigmaEval

SL · Reasoning / math / science · Rubric

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

Rank #14 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 58%
Percentile: 26.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

26.7% percentile inside its fair comparison set

58%Raw benchmark value

Professional reasoning34 benchmarks81.7%

APEX-Agents-AA

AA · Professional reasoning · Objective

Long-horizon agentic task completion.

Rank #12 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 28%
Percentile: 60.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `apexAgents`.

60.7% percentile inside its fair comparison set

28%Raw benchmark value

GDPval-AA

AA · Professional reasoning · Rubric

Agentic performance on economically valuable work tasks.

Rank #28 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 1,374
Percentile: 84.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gdpvalBreakdown.elo`.

84.1% percentile inside its fair comparison set

1,374Raw benchmark value

Legal Research Bench

VALS-AI · Professional reasoning · Objective

Applied legal research tasks.

Rank #3 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 38.5%
Percentile: 83.3%
Last updated: aging
Eligibility: headline eligible

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

83.3% percentile inside its fair comparison set

38.5%Raw benchmark valueCI 31.8% - 45.1%

SkillsBench

VALS-AI · Professional reasoning · Objective

Applied professional skills tasks.

Rank #9 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 49.1%
Percentile: 20%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: skillsbench; provider: Anthropic.

20% percentile inside its fair comparison set

49.1%Raw benchmark valueCI 40% - 58.1%

Public Benefits Bench

VALS-AI · Professional reasoning · Objective

Answering SNAP benefits questions across the public-benefits lifecycle.

Rank #4 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 62.4%
Percentile: 72.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: public-benefits-bench; provider: Anthropic.

72.7% percentile inside its fair comparison set

62.4%Raw benchmark valueCI 60% - 64.9%

Vals Index

VALS-AI · Professional reasoning · Combined

Weighted model performance across economically relevant Vals tasks.

Rank #7 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 60
Percentile: 76.9%
Last updated: aging
Eligibility: headline eligible

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

76.9% percentile inside its fair comparison set

60Raw benchmark valueCI 57 - 63

Harvey's Legal Agent Benchmark

VALS-AI · Professional reasoning · Objective

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

Rank #4 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 5%
Percentile: 76.9%
Last updated: aging
Eligibility: headline eligible

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

76.9% percentile inside its fair comparison set

5%Raw benchmark valueCI 1.8% - 8.2%

LegalBench

VALS-AI · Professional reasoning · Objective

Academic legal reasoning tasks.

Rank #43 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 82.1%
Percentile: 53.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: legal_bench; provider: Anthropic.

53.3% percentile inside its fair comparison set

82.1%Raw benchmark valueCI 81.2% - 83.1%

Finance Agent v2

VALS-AI · Professional reasoning · Objective

Core financial analyst tasks for agentic models.

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 51%
Percentile: 80%
Last updated: aging
Eligibility: headline eligible

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

80% percentile inside its fair comparison set

51%Raw benchmark valueCI 50.4% - 51.7%

TaxEval v2

VALS-AI · Professional reasoning · Objective

Answer quality on tax questions and responses.

Rank #2 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 77.1%
Percentile: 98.9%
Last updated: aging
Eligibility: headline eligible

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

98.9% percentile inside its fair comparison set

77.1%Raw benchmark valueCI 75.5% - 78.7%

MedCode

VALS-AI · Professional reasoning · Objective

Medical billing support and coding tasks.

Rank #38 · Source label: anthropic/claude-sonnet-4-20250514-thinking

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 35%
Percentile: 29.4%
Last updated: aging
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medcode; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

29.4% percentile inside its fair comparison set

35%Raw benchmark valueCI 31.2% - 38.8%

MedScribe

VALS-AI · Professional reasoning · Objective

Administrative documentation support for doctors.

Rank #40 · Source label: anthropic/claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 72.4%
Percentile: 24%
Last updated: aging
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medscribe; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

24% percentile inside its fair comparison set

72.4%Raw benchmark valueCI 68.6% - 76.2%

Text Arena · Expert

AR · Professional reasoning · Human

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

Rank #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,506
Percentile: 96.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: expert. Source rank: #15. Votes: 3658. Organization: anthropic. License: Proprietary.

96.4% percentile inside its fair comparison set

1,506Raw benchmark valueCI 1,496 - 1,517

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,483
Percentile: 97.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_business_and_management_and_financial_operations. Source rank: #12. Votes: 7734. Organization: anthropic. License: Proprietary.

97.5% percentile inside its fair comparison set

1,483Raw benchmark valueCI 1,475 - 1,491

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 #15

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,447
Percentile: 95.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_entertainment_and_sports_and_media. Source rank: #18. Votes: 8069. Organization: anthropic. License: Proprietary.

95.7% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,439 - 1,455

Text Arena · Industry Legal And Government

AR · Professional reasoning · Human

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

Rank #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,487
Percentile: 96.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_legal_and_government. Source rank: #13. Votes: 3019. Organization: anthropic. License: Proprietary.

96.6% percentile inside its fair comparison set

1,487Raw benchmark valueCI 1,475 - 1,499

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 #7

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,499
Percentile: 98.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_life_and_physical_and_social_science. Source rank: #9. Votes: 6405. Organization: anthropic. License: Proprietary.

98.1% percentile inside its fair comparison set

1,499Raw benchmark valueCI 1,491 - 1,507

Text Arena · Industry Mathematical

AR · Professional reasoning · Human

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

Rank #15

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,488
Percentile: 95.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_mathematical. Source rank: #18. Votes: 2169. Organization: anthropic. License: Proprietary.

95.5% percentile inside its fair comparison set

1,488Raw benchmark valueCI 1,474 - 1,501

Text Arena · Industry Medicine And Healthcare

AR · Professional reasoning · Human

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

Rank #19

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,487
Percentile: 93.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_medicine_and_healthcare. Source rank: #21. Votes: 2909. Organization: anthropic. License: Proprietary.

93.9% percentile inside its fair comparison set

1,487Raw benchmark valueCI 1,475 - 1,499

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,514
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_software_and_it_services. Source rank: #11. Votes: 15295. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,514Raw benchmark valueCI 1,508 - 1,520

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_writing_and_literature_and_language. Source rank: #23. Votes: 9327. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,449 - 1,463

Text Arena · Expert · No Style Control

AR · Professional reasoning · Human

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

Rank #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,498
Percentile: 96.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: expert. Source rank: #12. Votes: 3658. Organization: anthropic. License: Proprietary.

96.7% percentile inside its fair comparison set

1,498Raw benchmark valueCI 1,487 - 1,509

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,464
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_business_and_management_and_financial_operations. Source rank: #11. Votes: 7734. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,464Raw benchmark valueCI 1,457 - 1,472

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 #20

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,431
Percentile: 94.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_entertainment_and_sports_and_media. Source rank: #24. Votes: 8069. Organization: anthropic. License: Proprietary.

94.1% percentile inside its fair comparison set

1,431Raw benchmark valueCI 1,424 - 1,439

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 #16

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,471
Percentile: 95%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_legal_and_government. Source rank: #21. Votes: 3019. Organization: anthropic. License: Proprietary.

95% percentile inside its fair comparison set

1,471Raw benchmark valueCI 1,459 - 1,483

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 #17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,473
Percentile: 95%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_life_and_physical_and_social_science. Source rank: #21. Votes: 6405. Organization: anthropic. License: Proprietary.

95% percentile inside its fair comparison set

1,473Raw benchmark valueCI 1,465 - 1,481

Text Arena · Industry Mathematical · No Style Control

AR · Professional reasoning · Human

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

Rank #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,484
Percentile: 96.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_mathematical. Source rank: #13. Votes: 2169. Organization: anthropic. License: Proprietary.

96.8% percentile inside its fair comparison set

1,484Raw benchmark valueCI 1,470 - 1,498

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 #52

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,448
Percentile: 82.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_medicine_and_healthcare. Source rank: #60. Votes: 2909. Organization: anthropic. License: Proprietary.

82.7% percentile inside its fair comparison set

1,448Raw benchmark valueCI 1,436 - 1,460

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,490
Percentile: 97.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_software_and_it_services. Source rank: #12. Votes: 15295. Organization: anthropic. License: Proprietary.

97.5% percentile inside its fair comparison set

1,490Raw benchmark valueCI 1,484 - 1,496

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 #22

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,443
Percentile: 93.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_writing_and_literature_and_language. Source rank: #28. Votes: 9327. Organization: anthropic. License: Proprietary.

93.5% percentile inside its fair comparison set

1,443Raw benchmark valueCI 1,436 - 1,450

SAGE

VALS-AI · Professional reasoning · Objective

Student Assessment with Generative Evaluation.

Rank #18 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 46.6%
Percentile: 62.2%
Last updated: aging
Eligibility: headline eligible

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

62.2% percentile inside its fair comparison set

46.6%Raw benchmark valueCI 39.9% - 53.3%

Public Benefits Bench v1

VALS-AI · Professional reasoning · Objective

Answering public-benefits questions across the benefits lifecycle.

Rank #4 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 58.5%
Percentile: 75%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: public-benefits-bench-v1; provider: Anthropic.

75% percentile inside its fair comparison set

58.5%Raw benchmark valueCI 56% - 61%

Hallucination

BB · Professional reasoning · Rubric

Accuracy and fabrication resistance under prompts that invite unsupported claims.

Rank #9 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 76.6%
Percentile: 76.5%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-hallucination.

76.5% percentile inside its fair comparison set

76.6%Raw benchmark value

BS pushback

BB · Professional reasoning · Rubric

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

Rank #3 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 91.5%
Percentile: 88.9%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-pushback.

88.9% percentile inside its fair comparison set

91.5%Raw benchmark value

Search / tool use4 benchmarks62.8%

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 #87 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 75.7%
Percentile: 74.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `tau2`.

74.4% percentile inside its fair comparison set

75.7%Raw benchmark value

Search Arena

AR · Search / tool use · Human

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

Rank #8 · Source label: claude-sonnet-4-6-search

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,197
Percentile: 76.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6-search`. Category: overall. Source rank: #9. Votes: 81575. Organization: anthropic. License: Proprietary.

76.7% percentile inside its fair comparison set

1,197Raw benchmark valueCI 1,192 - 1,203

Search Arena · No Style Control

AR · Search / tool use · Human

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

Rank #6 · Source label: claude-sonnet-4-6-search

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,218
Percentile: 83.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6-search`. Category: overall. Source rank: #6. Votes: 81575. Organization: anthropic. License: Proprietary.

83.3% percentile inside its fair comparison set

1,218Raw benchmark valueCI 1,213 - 1,223

BrowseComp

OFF · Search / tool use · Objective

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

Rank #6

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 74%
Percentile: 16.7%
Last updated: stale
Eligibility: headline eligible

Verified against Anthropic's public Claude Sonnet 4.6 system card, checked on April 29, 2026. Uses the published BrowseComp figure. Checked 2026-04-29. Verification: manual_public_page_verification.

16.7% percentile inside its fair comparison set

74%Raw benchmark value

Long context2 benchmarks77.8%

Long Context Reasoning

AA · Long context · Objective

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

Rank #94 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 62.3%
Percentile: 76%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `lcr`.

76% percentile inside its fair comparison set

62.3%Raw benchmark value

CorpFin v2

VALS-AI · Long context · Objective

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

Rank #19 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 65.3%
Percentile: 79.5%
Last updated: aging
Eligibility: headline eligible

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

79.5% percentile inside its fair comparison set

65.3%Raw benchmark valueCI 63.5% - 67.1%

Vision understanding21 benchmarks69.7%

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 #58 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 69.2%
Percentile: 65.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `mmmuPro`.

65.2% percentile inside its fair comparison set

69.2%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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,278
Percentile: 89.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #15. Votes: 17336. Organization: anthropic. License: Proprietary.

89.9% percentile inside its fair comparison set

1,278Raw benchmark valueCI 1,271 - 1,285

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 #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,270
Percentile: 81.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing_vision. Source rank: #15. Votes: 949. Organization: anthropic. License: Proprietary.

81.8% percentile inside its fair comparison set

1,270Raw benchmark valueCI 1,250 - 1,291

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 #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,307
Percentile: 85.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: diagram. Source rank: #15. Votes: 4791. Organization: anthropic. License: Proprietary.

85.7% percentile inside its fair comparison set

1,307Raw benchmark valueCI 1,296 - 1,317

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 96.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #7. Votes: 7613. Organization: anthropic. License: Proprietary.

96.3% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,276 - 1,296

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,303
Percentile: 83.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: homework. Source rank: #16. Votes: 2662. Organization: anthropic. License: Proprietary.

83.8% percentile inside its fair comparison set

1,303Raw benchmark valueCI 1,290 - 1,316

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 #23

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,235
Percentile: 55.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: humor. Source rank: #29. Votes: 595. Organization: anthropic. License: Proprietary.

55.1% percentile inside its fair comparison set

1,235Raw benchmark valueCI 1,210 - 1,261

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,292
Percentile: 87.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: ocr. Source rank: #13. Votes: 12559. Organization: anthropic. License: Proprietary.

87.1% percentile inside its fair comparison set

1,292Raw benchmark valueCI 1,285 - 1,299

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,284
Percentile: 89.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #16. Votes: 17336. Organization: anthropic. License: Proprietary.

89.9% percentile inside its fair comparison set

1,284Raw benchmark valueCI 1,277 - 1,291

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,277
Percentile: 80%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing_vision. Source rank: #16. Votes: 949. Organization: anthropic. License: Proprietary.

80% percentile inside its fair comparison set

1,277Raw benchmark valueCI 1,256 - 1,297

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,306
Percentile: 88.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: diagram. Source rank: #13. Votes: 4791. Organization: anthropic. License: Proprietary.

88.6% percentile inside its fair comparison set

1,306Raw benchmark valueCI 1,296 - 1,316

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,300
Percentile: 96.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #7. Votes: 7613. Organization: anthropic. License: Proprietary.

96.3% percentile inside its fair comparison set

1,300Raw benchmark valueCI 1,291 - 1,310

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,311
Percentile: 83.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: homework. Source rank: #16. Votes: 2662. Organization: anthropic. License: Proprietary.

83.8% percentile inside its fair comparison set

1,311Raw benchmark valueCI 1,298 - 1,323

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 #25

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,243
Percentile: 51%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: humor. Source rank: #31. Votes: 595. Organization: anthropic. License: Proprietary.

51% percentile inside its fair comparison set

1,243Raw benchmark valueCI 1,218 - 1,269

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,298
Percentile: 87.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: ocr. Source rank: #13. Votes: 12559. Organization: anthropic. License: Proprietary.

87.1% percentile inside its fair comparison set

1,298Raw benchmark valueCI 1,291 - 1,305

MMMU Pro

VALS-AI · Vision understanding · Objective

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

Rank #16 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 83.6%
Percentile: 74.1%
Last updated: aging
Eligibility: headline eligible

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

74.1% percentile inside its fair comparison set

83.6%Raw benchmark valueCI 81.8% - 85.3%

VTB

SL · Vision understanding · Rubric

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

Rank #12 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 4.5%
Percentile: 18.2%
Last updated: stale
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

18.2% percentile inside its fair comparison set

4.5%Raw benchmark value

MMMU-Pro

OFF · Vision understanding · Objective

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

Rank #5

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 74.5%
Percentile: 20%
Last updated: stale
Eligibility: headline eligible

Verified against Anthropic's public Claude Sonnet 4.6 system card, checked on April 29, 2026. Uses the published no-tools MMMU-Pro figure. Checked 2026-04-29. Verification: manual_public_page_verification.

20% percentile inside its fair comparison set

74.5%Raw benchmark value

VISTA

SL · Vision understanding · Rubric

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

Rank #12 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 75%
Percentile: 35.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

35.7% percentile inside its fair comparison set

75%Raw benchmark value

Vision Arena · Creative Writing

AR · Vision understanding · Human

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

Rank #19 · Source label: claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,195
Percentile: 50%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Arena leaderboard dataset row `claude-sonnet-4-20250514`. Category: creative_writing. Source rank: #18. Votes: 156. Organization: anthropic. License: Proprietary. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

50% percentile inside its fair comparison set

1,195Raw benchmark valueCI 1,156 - 1,234

Vision Arena · Creative Writing · No Style Control

AR · Vision understanding · Human

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

Rank #21 · Source label: claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,191
Percentile: 43.8%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Arena leaderboard dataset row `claude-sonnet-4-20250514`. Category: creative_writing. Source rank: #20. Votes: 156. Organization: anthropic. License: Proprietary. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

43.8% percentile inside its fair comparison set

1,191Raw benchmark valueCI 1,152 - 1,231

Document understanding5 benchmarks69.5%

Document Arena

AR · Document understanding · Human

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

Rank #6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,482
Percentile: 79.2%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #8. Votes: 49424. Organization: anthropic. License: Proprietary.

79.2% percentile inside its fair comparison set

1,482Raw benchmark valueCI 1,476 - 1,488

Document Arena · No Style Control

AR · Document understanding · Human

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

Rank #4

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,487
Percentile: 87.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #6. Votes: 49424. Organization: anthropic. License: Proprietary.

87.5% percentile inside its fair comparison set

1,487Raw benchmark valueCI 1,481 - 1,493

Vals Multimodal Index

VALS-AI · Document understanding · Combined

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

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 61
Percentile: 73.7%
Last updated: aging
Eligibility: headline eligible

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

73.7% percentile inside its fair comparison set

61Raw benchmark valueCI 58 - 63

MortgageTax

VALS-AI · Document understanding · Objective

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

Rank #18 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 67.7%
Percentile: 71.7%
Last updated: aging
Eligibility: headline eligible

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

71.7% percentile inside its fair comparison set

67.7%Raw benchmark valueCI 65.9% - 69.5%

Multimodal mix

OC · Document understanding · Objective

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

Rank #12 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: OpenCompass
Raw value: 72.7%
Percentile: 35.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

35.7% percentile inside its fair comparison set

72.7%Raw benchmark value

Multilingual16 benchmarks88%

Text Arena · Chinese

AR · Multilingual · Human

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

Rank #25

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,503
Percentile: 91.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #30. Votes: 1919. Organization: anthropic. License: Proprietary.

91.9% percentile inside its fair comparison set

1,503Raw benchmark valueCI 1,489 - 1,518

Text Arena · French

AR · Multilingual · Human

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

Rank #27

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,478
Percentile: 88%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: french. Source rank: #35. Votes: 1159. Organization: anthropic. License: Proprietary.

88% percentile inside its fair comparison set

1,478Raw benchmark valueCI 1,459 - 1,498

Text Arena · German

AR · Multilingual · Human

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

Rank #21

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,460
Percentile: 91.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: german. Source rank: #27. Votes: 668. Organization: anthropic. License: Proprietary.

91.6% percentile inside its fair comparison set

1,460Raw benchmark valueCI 1,437 - 1,484

Text Arena · Japanese

AR · Multilingual · Human

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

Rank #26

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,423
Percentile: 87.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: japanese. Source rank: #36. Votes: 341. Organization: anthropic. License: Proprietary.

87.7% percentile inside its fair comparison set

1,423Raw benchmark valueCI 1,389 - 1,457

Text Arena · Korean

AR · Multilingual · Human

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

Rank #17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,431
Percentile: 92.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: korean. Source rank: #21. Votes: 577. Organization: anthropic. License: Proprietary.

92.3% percentile inside its fair comparison set

1,431Raw benchmark valueCI 1,405 - 1,457

Text Arena · Russian

AR · Multilingual · Human

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

Rank #22

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,463
Percentile: 92.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: russian. Source rank: #31. Votes: 4306. Organization: anthropic. License: Proprietary.

92.7% percentile inside its fair comparison set

1,463Raw benchmark valueCI 1,453 - 1,472

Text Arena · Spanish

AR · Multilingual · Human

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

Rank #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,465
Percentile: 92.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: spanish. Source rank: #21. Votes: 1206. Organization: anthropic. License: Proprietary.

92.1% percentile inside its fair comparison set

1,465Raw benchmark valueCI 1,446 - 1,484

Text Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #31

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,483
Percentile: 89.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #36. Votes: 1919. Organization: anthropic. License: Proprietary.

89.8% percentile inside its fair comparison set

1,483Raw benchmark valueCI 1,469 - 1,498

Text Arena · French · No Style Control

AR · Multilingual · Human

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

Rank #32

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,461
Percentile: 85.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: french. Source rank: #38. Votes: 1159. Organization: anthropic. License: Proprietary.

85.6% percentile inside its fair comparison set

1,461Raw benchmark valueCI 1,442 - 1,480

Text Arena · German · No Style Control

AR · Multilingual · Human

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

Rank #37

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,432
Percentile: 84.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: german. Source rank: #46. Votes: 668. Organization: anthropic. License: Proprietary.

84.8% percentile inside its fair comparison set

1,432Raw benchmark valueCI 1,408 - 1,455

Text Arena · Japanese · No Style Control

AR · Multilingual · Human

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

Rank #32

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,402
Percentile: 84.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: japanese. Source rank: #42. Votes: 341. Organization: anthropic. License: Proprietary.

84.7% percentile inside its fair comparison set

1,402Raw benchmark valueCI 1,368 - 1,436

Text Arena · Korean · No Style Control

AR · Multilingual · Human

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

Rank #23

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,412
Percentile: 89.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: korean. Source rank: #29. Votes: 577. Organization: anthropic. License: Proprietary.

89.4% percentile inside its fair comparison set

1,412Raw benchmark valueCI 1,386 - 1,438

Text Arena · Russian · No Style Control

AR · Multilingual · Human

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

Rank #32

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 89.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: russian. Source rank: #40. Votes: 4306. Organization: anthropic. License: Proprietary.

89.3% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,427 - 1,447

Text Arena · Spanish · No Style Control

AR · Multilingual · Human

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

Rank #16

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,460
Percentile: 93%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: spanish. Source rank: #18. Votes: 1206. Organization: anthropic. License: Proprietary.

93% percentile inside its fair comparison set

1,460Raw benchmark valueCI 1,442 - 1,479

Vision Arena · Chinese

AR · Multilingual · Human

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

Rank #15

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,302
Percentile: 81.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #20. Votes: 997. Organization: anthropic. License: Proprietary.

81.8% percentile inside its fair comparison set

1,302Raw benchmark valueCI 1,277 - 1,326

Vision Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #21

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,305
Percentile: 74%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #26. Votes: 997. Organization: anthropic. License: Proprietary.

74% percentile inside its fair comparison set

1,305Raw benchmark valueCI 1,281 - 1,329

Source links and registry checks

official

Anthropic model overview

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Arena

Aug 19, 2026

source →

official

Artificial Analysis

Aug 19, 2026

source →

Model profile · Anthropic

Claude Sonnet 4.6

Closed weightspremium · registry tag 2026 workhorse

Thin verified coverage

Reads as thin verified coverage across the resolved source data.

Visible coverage: 52.1%
Verified coverage: 47.3%
Spread: 97.8%
Last verified: Aug 19, 2026

66%bench fit

textcodevisiondocumentsearch10 aliases47 official source links

Open compare

Data version

Current snapshot.

Published Aug 19, 2026Model list checked2 stale sources216 claim warnings9 providers · 1149 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.

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

Chat / text28 benchmarks74.4%

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 #66 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 35
Percentile: 86.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `intelligenceIndex`.

86.1% percentile inside its fair comparison set

35Raw 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 #42 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 36.5%
Percentile: 88.8%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceAccuracy`.

88.8% percentile inside its fair comparison set

36.5%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 #99 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 31.5%
Percentile: 73.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `omniscienceNonHallucination`.

73.2% percentile inside its fair comparison set

31.5%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 #179 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 41.2%
Percentile: 48.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `ifbench`.

48.4% percentile inside its fair comparison set

41.2%Raw benchmark value

Blended 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 #266 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $6 /1M tokens
Percentile: 12.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mBlended0To3To1`.

12.3% percentile inside its fair comparison set

$6 /1M tokensRaw 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 #266 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $3 /1M input tokens
Percentile: 12.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mInputTokens`.

12.3% percentile inside its fair comparison set

$3 /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 #267 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: $15 /1M output tokens
Percentile: 12.3%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `price1mOutputTokens`.

12.3% percentile inside its fair comparison set

$15 /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 #197 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 43.2 tokens/s
Percentile: 12.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianOutputTokensPerSecond`.

12.5% percentile inside its fair comparison set

43.2 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 #220 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 149.24s
Percentile: 2.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstTokenSeconds`.

2.2% percentile inside its fair comparison set

149.24sRaw 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 #220 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 149.24s
Percentile: 2.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `medianTimeToFirstAnswerTokenSeconds`.

2.2% percentile inside its fair comparison set

149.24sRaw 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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,472
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #24. Votes: 39561. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,472Raw benchmark valueCI 1,468 - 1,476

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 #20

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,446
Percentile: 94.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing. Source rank: #28. Votes: 6301. Organization: anthropic. License: Proprietary.

94.1% percentile inside its fair comparison set

1,446Raw benchmark valueCI 1,438 - 1,455

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,485
Percentile: 97.2%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #12. Votes: 18942. Organization: anthropic. License: Proprietary.

97.2% percentile inside its fair comparison set

1,485Raw benchmark valueCI 1,479 - 1,490

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,481
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: exclude_ties. Source rank: #23. Votes: 29825. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,481Raw benchmark valueCI 1,476 - 1,487

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,504
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts. Source rank: #10. Votes: 25042. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,504Raw benchmark valueCI 1,499 - 1,509

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 #6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,509
Percentile: 98.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts_english. Source rank: #8. Votes: 12541. Organization: anthropic. License: Proprietary.

98.5% percentile inside its fair comparison set

1,509Raw benchmark valueCI 1,503 - 1,516

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,479
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: instruction_following. Source rank: #10. Votes: 12635. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,479Raw benchmark valueCI 1,473 - 1,485

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,495
Percentile: 97.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: longer_query. Source rank: #11. Votes: 15661. Organization: anthropic. License: Proprietary.

97.7% percentile inside its fair comparison set

1,495Raw benchmark valueCI 1,488 - 1,501

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 #14

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,484
Percentile: 96%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: multi_turn. Source rank: #16. Votes: 7169. Organization: anthropic. License: Proprietary.

96% percentile inside its fair comparison set

1,484Raw benchmark valueCI 1,477 - 1,492

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #22. Votes: 39561. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,452 - 1,461

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 #26

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,432
Percentile: 92.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing. Source rank: #34. Votes: 6301. Organization: anthropic. License: Proprietary.

92.3% percentile inside its fair comparison set

1,432Raw benchmark valueCI 1,424 - 1,440

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 #13

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,472
Percentile: 96.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #15. Votes: 18942. Organization: anthropic. License: Proprietary.

96.3% percentile inside its fair comparison set

1,472Raw benchmark valueCI 1,467 - 1,478

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: exclude_ties. Source rank: #22. Votes: 29825. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,451 - 1,462

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,484
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts. Source rank: #10. Votes: 25042. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,484Raw benchmark valueCI 1,479 - 1,489

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,494
Percentile: 98.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: hard_prompts_english. Source rank: #7. Votes: 12541. Organization: anthropic. License: Proprietary.

98.8% percentile inside its fair comparison set

1,494Raw benchmark valueCI 1,488 - 1,501

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,479
Percentile: 98.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: instruction_following. Source rank: #7. Votes: 12635. Organization: anthropic. License: Proprietary.

98.8% percentile inside its fair comparison set

1,479Raw benchmark valueCI 1,472 - 1,485

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,481
Percentile: 97%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: longer_query. Source rank: #13. Votes: 15661. Organization: anthropic. License: Proprietary.

97% percentile inside its fair comparison set

1,481Raw benchmark valueCI 1,475 - 1,487

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,466
Percentile: 94.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: multi_turn. Source rank: #22. Votes: 7169. Organization: anthropic. License: Proprietary.

94.7% percentile inside its fair comparison set

1,466Raw benchmark valueCI 1,458 - 1,474

Coding30 benchmarks77%

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 #19 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 42.4%
Percentile: 94.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `terminalbenchHard`.

94.5% percentile inside its fair comparison set

42.4%Raw benchmark value

SciCode

AA · Coding · Objective

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

Rank #52 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 44.1%
Percentile: 88.5%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `scicode`.

88.5% percentile inside its fair comparison set

44.1%Raw benchmark value

Coding Index

AA · Coding · Combined

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

Rank #31 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 63
Percentile: 84%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `codingIndex`.

84% percentile inside its fair comparison set

63Raw benchmark value

Agentic Index

AA · Coding · Combined

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

Rank #21 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 42
Percentile: 85%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `agenticIndex`.

85% percentile inside its fair comparison set

42Raw benchmark value

Code 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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,521
Percentile: 90.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #12. Votes: 13612. Organization: anthropic. License: Proprietary.

90.4% percentile inside its fair comparison set

1,521Raw benchmark valueCI 1,515 - 1,527

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,521
Percentile: 90.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: webdev. Source rank: #12. Votes: 13612. Organization: anthropic. License: Proprietary.

90.4% percentile inside its fair comparison set

1,521Raw benchmark valueCI 1,515 - 1,527

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,515
Percentile: 89%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: webdev-html. Source rank: #12. Votes: 1879. Organization: anthropic. License: Proprietary.

89% percentile inside its fair comparison set

1,515Raw benchmark valueCI 1,501 - 1,529

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,515
Percentile: 88.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: webdev-react. Source rank: #12. Votes: 11688. Organization: anthropic. License: Proprietary.

88.1% percentile inside its fair comparison set

1,515Raw benchmark valueCI 1,509 - 1,522

Code Migration

VALS-AI · Coding · Objective

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

Rank #5 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 39.9%
Percentile: 81.8%
Last updated: aging
Eligibility: headline eligible

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

81.8% percentile inside its fair comparison set

39.9%Raw benchmark valueCI 31.8% - 48%

LiveCodeBench

VALS-AI · Coding · Objective

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

Rank #34 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 82.1%
Percentile: 63.3%
Last updated: aging
Eligibility: headline eligible

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

63.3% percentile inside its fair comparison set

82.1%Raw benchmark valueCI 80% - 84.2%

ProgramBench

VALS-AI · Coding · Objective

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

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 0.5%
Percentile: 90%
Last updated: aging
Eligibility: headline eligible

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

90% percentile inside its fair comparison set

0.5%Raw benchmark valueCI 0% - 1.5%

SWE-bench Verified

VALS-AI · Coding · Objective

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

Rank #12 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 77.4%
Percentile: 79.6%
Last updated: aging
Eligibility: headline eligible

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

79.6% percentile inside its fair comparison set

77.4%Raw benchmark valueCI 73.7% - 81.1%

Terminal-Bench 2.1

VALS-AI · Coding · Objective

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

Rank #11 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 57.3%
Percentile: 63%
Last updated: aging
Eligibility: headline eligible

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

63% percentile inside its fair comparison set

57.3%Raw benchmark valueCI 56% - 58.6%

Vibe Code Bench v1.1

VALS-AI · Coding · Objective

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

Rank #9 · Source label: anthropic/claude-sonnet-4-6-claude-code

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 55.8%
Percentile: 83.7%
Last updated: aging
Eligibility: headline eligible

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

83.7% percentile inside its fair comparison set

55.8%Raw benchmark valueCI 46.6% - 65%

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,527
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: coding. Source rank: #10. Votes: 10802. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,527Raw benchmark valueCI 1,520 - 1,534

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,502
Percentile: 98.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: coding. Source rank: #7. Votes: 10802. Organization: anthropic. License: Proprietary.

98.8% percentile inside its fair comparison set

1,502Raw benchmark valueCI 1,495 - 1,508

IOI

VALS-AI · Coding · Objective

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

Rank #31 · Source label: anthropic/claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 6.5%
Percentile: 34.1%
Last updated: aging
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: ioi; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

34.1% percentile inside its fair comparison set

6.5%Raw benchmark valueCI 2.4% - 10.6%

Code Arena · Image To Webdev

AR · Coding · Human

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

Rank #2

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,557
Percentile: 94.1%
Last updated: stale
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: image_to_webdev. Source rank: #2. Votes: 3158. Organization: anthropic. License: Proprietary.

94.1% percentile inside its fair comparison set

1,557Raw benchmark valueCI 1,544 - 1,570

Debugging

BB · Coding · Rubric

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

Rank #3 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 86.6%
Percentile: 91.7%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-debugging.

91.7% percentile inside its fair comparison set

86.6%Raw benchmark value

Refactoring

BB · Coding · Rubric

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

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 69.4%
Percentile: 76.2%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-refactoring.

76.2% percentile inside its fair comparison set

69.4%Raw benchmark value

Security

BB · Coding · Rubric

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

Rank #2 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 85.3%
Percentile: 100%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-security.

100% percentile inside its fair comparison set

85.3%Raw benchmark value

UI bench

BB · Coding · Rubric

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

Rank #1 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 81.5%
Percentile: 100%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ui.

100% percentile inside its fair comparison set

81.5%Raw benchmark value

Speed throughput

BB · Coding · Speed / cost

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

Rank #13 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 95.3 t/s
Percentile: 52%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-throughput.

52% percentile inside its fair comparison set

95.3 t/sRaw benchmark value

Speed TTFT

BB · Coding · Speed / cost

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

Rank #9 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 1207.00ms
Percentile: 68%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-ttft.

68% percentile inside its fair comparison set

1207.00msRaw benchmark value

Speed cost

BB · Coding · Speed / cost

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

Rank #22 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: $0.5 $
Percentile: 16%
Last updated: stale
Eligibility: headline eligible

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

16% percentile inside its fair comparison set

$0.5 $Raw benchmark value

Terminal-Bench 2.0

OFF · Coding · Objective

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

Rank #6

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 59.1%
Percentile: 16.7%
Last updated: stale
Eligibility: headline eligible

16.7% percentile inside its fair comparison set

59.1%Raw benchmark value

SWE-Bench Verified

OFF · Coding · Objective

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

Rank #4

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 79.6%
Percentile: 40%
Last updated: stale
Eligibility: headline eligible

40% percentile inside its fair comparison set

79.6%Raw benchmark value

Overall quality

BB · Coding · Combined

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

Rank #5 · Source label: Claude Sonnet 4.6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 89.5%
Percentile: 81.8%
Last updated: archived
Eligibility: headline eligible

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

81.8% percentile inside its fair comparison set

89.5%Raw benchmark value

Overall vibe

BB · Coding · Combined

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

Rank #7 · Source label: Claude Sonnet 4.6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 82.7%
Percentile: 72.7%
Last updated: archived
Eligibility: headline eligible

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

72.7% percentile inside its fair comparison set

82.7%Raw benchmark value

Overall coverage

BB · Coding · Combined

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

Rank #6 · Source label: Claude Sonnet 4.6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

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

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

100% percentile inside its fair comparison set

100%Raw benchmark value

Reasoning / math / science11 benchmarks65.6%

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 #133 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 11.2%
Percentile: 70.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `hle`.

70.1% percentile inside its fair comparison set

11.2%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 #100 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 79.7%
Percentile: 78%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gpqa`.

78% percentile inside its fair comparison set

79.7%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 #86 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 0.9%
Percentile: 77%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `critpt`.

77% percentile inside its fair comparison set

0.9%Raw benchmark value

ProofBench

VALS-AI · Reasoning / math / science · Objective

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

Rank #7 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 45%
Percentile: 82.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: proof_bench; provider: Anthropic.

82.9% percentile inside its fair comparison set

45%Raw benchmark valueCI 35.2% - 54.8%

GPQA Diamond

VALS-AI · Reasoning / math / science · Objective

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

Rank #28 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 85.6%
Percentile: 71.9%
Last updated: aging
Eligibility: headline eligible

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

71.9% percentile inside its fair comparison set

85.6%Raw benchmark valueCI 80.9% - 90.3%

MMLU Pro

VALS-AI · Reasoning / math / science · Objective

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

Rank #16 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 87.3%
Percentile: 83.1%
Last updated: aging
Eligibility: headline eligible

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

83.1% percentile inside its fair comparison set

87.3%Raw benchmark valueCI 86.5% - 88.2%

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 #27

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,459
Percentile: 91.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: math. Source rank: #33. Votes: 2296. Organization: anthropic. License: Proprietary.

91.7% percentile inside its fair comparison set

1,459Raw benchmark valueCI 1,446 - 1,471

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 #25

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 92.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: math. Source rank: #31. Votes: 2296. Organization: anthropic. License: Proprietary.

92.4% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,443 - 1,469

Reasoning

BB · Reasoning / math / science · Rubric

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

Rank #15 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 37.2%
Percentile: 33.3%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-reasoning.

33.3% percentile inside its fair comparison set

37.2%Raw benchmark value

Humanity's Last Exam

OFF · Reasoning / math / science · Objective

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

Rank #7

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 33.2%
Percentile: 14.3%
Last updated: stale
Eligibility: headline eligible

14.3% percentile inside its fair comparison set

33.2%Raw benchmark value

EnigmaEval

SL · Reasoning / math / science · Rubric

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

Rank #14 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 58%
Percentile: 26.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

26.7% percentile inside its fair comparison set

58%Raw benchmark value

Professional reasoning34 benchmarks81.7%

APEX-Agents-AA

AA · Professional reasoning · Objective

Long-horizon agentic task completion.

Rank #12 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 28%
Percentile: 60.7%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `apexAgents`.

60.7% percentile inside its fair comparison set

28%Raw benchmark value

GDPval-AA

AA · Professional reasoning · Rubric

Agentic performance on economically valuable work tasks.

Rank #28 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 1,374
Percentile: 84.1%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `gdpvalBreakdown.elo`.

84.1% percentile inside its fair comparison set

1,374Raw benchmark value

Legal Research Bench

VALS-AI · Professional reasoning · Objective

Applied legal research tasks.

Rank #3 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 38.5%
Percentile: 83.3%
Last updated: aging
Eligibility: headline eligible

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

83.3% percentile inside its fair comparison set

38.5%Raw benchmark valueCI 31.8% - 45.1%

SkillsBench

VALS-AI · Professional reasoning · Objective

Applied professional skills tasks.

Rank #9 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 49.1%
Percentile: 20%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: skillsbench; provider: Anthropic.

20% percentile inside its fair comparison set

49.1%Raw benchmark valueCI 40% - 58.1%

Public Benefits Bench

VALS-AI · Professional reasoning · Objective

Answering SNAP benefits questions across the public-benefits lifecycle.

Rank #4 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 62.4%
Percentile: 72.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: public-benefits-bench; provider: Anthropic.

72.7% percentile inside its fair comparison set

62.4%Raw benchmark valueCI 60% - 64.9%

Vals Index

VALS-AI · Professional reasoning · Combined

Weighted model performance across economically relevant Vals tasks.

Rank #7 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 60
Percentile: 76.9%
Last updated: aging
Eligibility: headline eligible

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

76.9% percentile inside its fair comparison set

60Raw benchmark valueCI 57 - 63

Harvey's Legal Agent Benchmark

VALS-AI · Professional reasoning · Objective

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

Rank #4 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 5%
Percentile: 76.9%
Last updated: aging
Eligibility: headline eligible

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

76.9% percentile inside its fair comparison set

5%Raw benchmark valueCI 1.8% - 8.2%

LegalBench

VALS-AI · Professional reasoning · Objective

Academic legal reasoning tasks.

Rank #43 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 82.1%
Percentile: 53.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: legal_bench; provider: Anthropic.

53.3% percentile inside its fair comparison set

82.1%Raw benchmark valueCI 81.2% - 83.1%

Finance Agent v2

VALS-AI · Professional reasoning · Objective

Core financial analyst tasks for agentic models.

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 51%
Percentile: 80%
Last updated: aging
Eligibility: headline eligible

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

80% percentile inside its fair comparison set

51%Raw benchmark valueCI 50.4% - 51.7%

TaxEval v2

VALS-AI · Professional reasoning · Objective

Answer quality on tax questions and responses.

Rank #2 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 77.1%
Percentile: 98.9%
Last updated: aging
Eligibility: headline eligible

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

98.9% percentile inside its fair comparison set

77.1%Raw benchmark valueCI 75.5% - 78.7%

MedCode

VALS-AI · Professional reasoning · Objective

Medical billing support and coding tasks.

Rank #38 · Source label: anthropic/claude-sonnet-4-20250514-thinking

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 35%
Percentile: 29.4%
Last updated: aging
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medcode; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

29.4% percentile inside its fair comparison set

35%Raw benchmark valueCI 31.2% - 38.8%

MedScribe

VALS-AI · Professional reasoning · Objective

Administrative documentation support for doctors.

Rank #40 · Source label: anthropic/claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 72.4%
Percentile: 24%
Last updated: aging
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Parsed from Vals AI BenchmarkView overall scores. Vals slug: medscribe; provider: Anthropic. Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

24% percentile inside its fair comparison set

72.4%Raw benchmark valueCI 68.6% - 76.2%

Text Arena · Expert

AR · Professional reasoning · Human

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

Rank #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,506
Percentile: 96.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: expert. Source rank: #15. Votes: 3658. Organization: anthropic. License: Proprietary.

96.4% percentile inside its fair comparison set

1,506Raw benchmark valueCI 1,496 - 1,517

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,483
Percentile: 97.5%
Last updated: aging
Eligibility: headline eligible

97.5% percentile inside its fair comparison set

1,483Raw benchmark valueCI 1,475 - 1,491

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 #15

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,447
Percentile: 95.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_entertainment_and_sports_and_media. Source rank: #18. Votes: 8069. Organization: anthropic. License: Proprietary.

95.7% percentile inside its fair comparison set

1,447Raw benchmark valueCI 1,439 - 1,455

Text Arena · Industry Legal And Government

AR · Professional reasoning · Human

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

Rank #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,487
Percentile: 96.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_legal_and_government. Source rank: #13. Votes: 3019. Organization: anthropic. License: Proprietary.

96.6% percentile inside its fair comparison set

1,487Raw benchmark valueCI 1,475 - 1,499

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 #7

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,499
Percentile: 98.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_life_and_physical_and_social_science. Source rank: #9. Votes: 6405. Organization: anthropic. License: Proprietary.

98.1% percentile inside its fair comparison set

1,499Raw benchmark valueCI 1,491 - 1,507

Text Arena · Industry Mathematical

AR · Professional reasoning · Human

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

Rank #15

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,488
Percentile: 95.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_mathematical. Source rank: #18. Votes: 2169. Organization: anthropic. License: Proprietary.

95.5% percentile inside its fair comparison set

1,488Raw benchmark valueCI 1,474 - 1,501

Text Arena · Industry Medicine And Healthcare

AR · Professional reasoning · Human

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

Rank #19

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,487
Percentile: 93.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_medicine_and_healthcare. Source rank: #21. Votes: 2909. Organization: anthropic. License: Proprietary.

93.9% percentile inside its fair comparison set

1,487Raw benchmark valueCI 1,475 - 1,499

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,514
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_software_and_it_services. Source rank: #11. Votes: 15295. Organization: anthropic. License: Proprietary.

97.8% percentile inside its fair comparison set

1,514Raw benchmark valueCI 1,508 - 1,520

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 #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,456
Percentile: 94.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_writing_and_literature_and_language. Source rank: #23. Votes: 9327. Organization: anthropic. License: Proprietary.

94.8% percentile inside its fair comparison set

1,456Raw benchmark valueCI 1,449 - 1,463

Text Arena · Expert · No Style Control

AR · Professional reasoning · Human

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

Rank #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,498
Percentile: 96.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: expert. Source rank: #12. Votes: 3658. Organization: anthropic. License: Proprietary.

96.7% percentile inside its fair comparison set

1,498Raw benchmark valueCI 1,487 - 1,509

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 #8

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,464
Percentile: 97.8%
Last updated: aging
Eligibility: headline eligible

97.8% percentile inside its fair comparison set

1,464Raw benchmark valueCI 1,457 - 1,472

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 #20

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,431
Percentile: 94.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_entertainment_and_sports_and_media. Source rank: #24. Votes: 8069. Organization: anthropic. License: Proprietary.

94.1% percentile inside its fair comparison set

1,431Raw benchmark valueCI 1,424 - 1,439

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 #16

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,471
Percentile: 95%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_legal_and_government. Source rank: #21. Votes: 3019. Organization: anthropic. License: Proprietary.

95% percentile inside its fair comparison set

1,471Raw benchmark valueCI 1,459 - 1,483

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 #17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,473
Percentile: 95%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_life_and_physical_and_social_science. Source rank: #21. Votes: 6405. Organization: anthropic. License: Proprietary.

95% percentile inside its fair comparison set

1,473Raw benchmark valueCI 1,465 - 1,481

Text Arena · Industry Mathematical · No Style Control

AR · Professional reasoning · Human

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

Rank #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,484
Percentile: 96.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_mathematical. Source rank: #13. Votes: 2169. Organization: anthropic. License: Proprietary.

96.8% percentile inside its fair comparison set

1,484Raw benchmark valueCI 1,470 - 1,498

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 #52

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,448
Percentile: 82.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_medicine_and_healthcare. Source rank: #60. Votes: 2909. Organization: anthropic. License: Proprietary.

82.7% percentile inside its fair comparison set

1,448Raw benchmark valueCI 1,436 - 1,460

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,490
Percentile: 97.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_software_and_it_services. Source rank: #12. Votes: 15295. Organization: anthropic. License: Proprietary.

97.5% percentile inside its fair comparison set

1,490Raw benchmark valueCI 1,484 - 1,496

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 #22

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,443
Percentile: 93.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: industry_writing_and_literature_and_language. Source rank: #28. Votes: 9327. Organization: anthropic. License: Proprietary.

93.5% percentile inside its fair comparison set

1,443Raw benchmark valueCI 1,436 - 1,450

SAGE

VALS-AI · Professional reasoning · Objective

Student Assessment with Generative Evaluation.

Rank #18 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 46.6%
Percentile: 62.2%
Last updated: aging
Eligibility: headline eligible

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

62.2% percentile inside its fair comparison set

46.6%Raw benchmark valueCI 39.9% - 53.3%

Public Benefits Bench v1

VALS-AI · Professional reasoning · Objective

Answering public-benefits questions across the benefits lifecycle.

Rank #4 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 58.5%
Percentile: 75%
Last updated: aging
Eligibility: headline eligible

Parsed from Vals AI BenchmarkView overall scores. Vals slug: public-benefits-bench-v1; provider: Anthropic.

75% percentile inside its fair comparison set

58.5%Raw benchmark valueCI 56% - 61%

Hallucination

BB · Professional reasoning · Rubric

Accuracy and fabrication resistance under prompts that invite unsupported claims.

Rank #9 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 76.6%
Percentile: 76.5%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-hallucination.

76.5% percentile inside its fair comparison set

76.6%Raw benchmark value

BS pushback

BB · Professional reasoning · Rubric

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

Rank #3 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: BridgeBench
Raw value: 91.5%
Percentile: 88.9%
Last updated: stale
Eligibility: headline eligible

Parsed from the BridgeBench page for bridgebench-pushback.

88.9% percentile inside its fair comparison set

91.5%Raw benchmark value

Search / tool use4 benchmarks62.8%

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 #87 · Source label: Claude Sonnet 4.6 (Adaptive Reasoning, Max Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 75.7%
Percentile: 74.4%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `tau2`.

74.4% percentile inside its fair comparison set

75.7%Raw benchmark value

Search Arena

AR · Search / tool use · Human

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

Rank #8 · Source label: claude-sonnet-4-6-search

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,197
Percentile: 76.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6-search`. Category: overall. Source rank: #9. Votes: 81575. Organization: anthropic. License: Proprietary.

76.7% percentile inside its fair comparison set

1,197Raw benchmark valueCI 1,192 - 1,203

Search Arena · No Style Control

AR · Search / tool use · Human

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

Rank #6 · Source label: claude-sonnet-4-6-search

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,218
Percentile: 83.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6-search`. Category: overall. Source rank: #6. Votes: 81575. Organization: anthropic. License: Proprietary.

83.3% percentile inside its fair comparison set

1,218Raw benchmark valueCI 1,213 - 1,223

BrowseComp

OFF · Search / tool use · Objective

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

Rank #6

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 74%
Percentile: 16.7%
Last updated: stale
Eligibility: headline eligible

Verified against Anthropic's public Claude Sonnet 4.6 system card, checked on April 29, 2026. Uses the published BrowseComp figure. Checked 2026-04-29. Verification: manual_public_page_verification.

16.7% percentile inside its fair comparison set

74%Raw benchmark value

Long context2 benchmarks77.8%

Long Context Reasoning

AA · Long context · Objective

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

Rank #94 · Source label: Claude Sonnet 4.6 (Non-reasoning, High Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 62.3%
Percentile: 76%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `lcr`.

76% percentile inside its fair comparison set

62.3%Raw benchmark value

CorpFin v2

VALS-AI · Long context · Objective

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

Rank #19 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 65.3%
Percentile: 79.5%
Last updated: aging
Eligibility: headline eligible

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

79.5% percentile inside its fair comparison set

65.3%Raw benchmark valueCI 63.5% - 67.1%

Vision understanding21 benchmarks69.7%

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 #58 · Source label: Claude Sonnet 4.6 (Non-reasoning, Low Effort)

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Artificial Analysis
Raw value: 69.2%
Percentile: 65.2%
Last updated: recent
Eligibility: headline eligible

Parsed from Artificial Analysis public leaderboard field `mmmuPro`.

65.2% percentile inside its fair comparison set

69.2%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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,278
Percentile: 89.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #15. Votes: 17336. Organization: anthropic. License: Proprietary.

89.9% percentile inside its fair comparison set

1,278Raw benchmark valueCI 1,271 - 1,285

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 #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,270
Percentile: 81.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing_vision. Source rank: #15. Votes: 949. Organization: anthropic. License: Proprietary.

81.8% percentile inside its fair comparison set

1,270Raw benchmark valueCI 1,250 - 1,291

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 #11

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,307
Percentile: 85.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: diagram. Source rank: #15. Votes: 4791. Organization: anthropic. License: Proprietary.

85.7% percentile inside its fair comparison set

1,307Raw benchmark valueCI 1,296 - 1,317

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,286
Percentile: 96.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #7. Votes: 7613. Organization: anthropic. License: Proprietary.

96.3% percentile inside its fair comparison set

1,286Raw benchmark valueCI 1,276 - 1,296

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,303
Percentile: 83.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: homework. Source rank: #16. Votes: 2662. Organization: anthropic. License: Proprietary.

83.8% percentile inside its fair comparison set

1,303Raw benchmark valueCI 1,290 - 1,316

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 #23

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,235
Percentile: 55.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: humor. Source rank: #29. Votes: 595. Organization: anthropic. License: Proprietary.

55.1% percentile inside its fair comparison set

1,235Raw benchmark valueCI 1,210 - 1,261

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,292
Percentile: 87.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: ocr. Source rank: #13. Votes: 12559. Organization: anthropic. License: Proprietary.

87.1% percentile inside its fair comparison set

1,292Raw benchmark valueCI 1,285 - 1,299

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,284
Percentile: 89.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #16. Votes: 17336. Organization: anthropic. License: Proprietary.

89.9% percentile inside its fair comparison set

1,284Raw benchmark valueCI 1,277 - 1,291

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,277
Percentile: 80%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: creative_writing_vision. Source rank: #16. Votes: 949. Organization: anthropic. License: Proprietary.

80% percentile inside its fair comparison set

1,277Raw benchmark valueCI 1,256 - 1,297

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 #9

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,306
Percentile: 88.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: diagram. Source rank: #13. Votes: 4791. Organization: anthropic. License: Proprietary.

88.6% percentile inside its fair comparison set

1,306Raw benchmark valueCI 1,296 - 1,316

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 #5

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,300
Percentile: 96.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: english. Source rank: #7. Votes: 7613. Organization: anthropic. License: Proprietary.

96.3% percentile inside its fair comparison set

1,300Raw benchmark valueCI 1,291 - 1,310

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 #12

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,311
Percentile: 83.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: homework. Source rank: #16. Votes: 2662. Organization: anthropic. License: Proprietary.

83.8% percentile inside its fair comparison set

1,311Raw benchmark valueCI 1,298 - 1,323

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 #25

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,243
Percentile: 51%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: humor. Source rank: #31. Votes: 595. Organization: anthropic. License: Proprietary.

51% percentile inside its fair comparison set

1,243Raw benchmark valueCI 1,218 - 1,269

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 #10

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,298
Percentile: 87.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: ocr. Source rank: #13. Votes: 12559. Organization: anthropic. License: Proprietary.

87.1% percentile inside its fair comparison set

1,298Raw benchmark valueCI 1,291 - 1,305

MMMU Pro

VALS-AI · Vision understanding · Objective

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

Rank #16 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 83.6%
Percentile: 74.1%
Last updated: aging
Eligibility: headline eligible

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

74.1% percentile inside its fair comparison set

83.6%Raw benchmark valueCI 81.8% - 85.3%

VTB

SL · Vision understanding · Rubric

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

Rank #12 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 4.5%
Percentile: 18.2%
Last updated: stale
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

18.2% percentile inside its fair comparison set

4.5%Raw benchmark value

MMMU-Pro

OFF · Vision understanding · Objective

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

Rank #5

Official company resultmanual verifiedmanual verified

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Provider official evals
Raw value: 74.5%
Percentile: 20%
Last updated: stale
Eligibility: headline eligible

20% percentile inside its fair comparison set

74.5%Raw benchmark value

VISTA

SL · Vision understanding · Rubric

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

Rank #12 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Scale Labs
Raw value: 75%
Percentile: 35.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

35.7% percentile inside its fair comparison set

75%Raw benchmark value

Vision Arena · Creative Writing

AR · Vision understanding · Human

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

Rank #19 · Source label: claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,195
Percentile: 50%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

50% percentile inside its fair comparison set

1,195Raw benchmark valueCI 1,156 - 1,234

Vision Arena · Creative Writing · No Style Control

AR · Vision understanding · Human

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

Rank #21 · Source label: claude-sonnet-4-20250514

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,191
Percentile: 43.8%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

43.8% percentile inside its fair comparison set

1,191Raw benchmark valueCI 1,152 - 1,231

Document understanding5 benchmarks69.5%

Document Arena

AR · Document understanding · Human

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

Rank #6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,482
Percentile: 79.2%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #8. Votes: 49424. Organization: anthropic. License: Proprietary.

79.2% percentile inside its fair comparison set

1,482Raw benchmark valueCI 1,476 - 1,488

Document Arena · No Style Control

AR · Document understanding · Human

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

Rank #4

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,487
Percentile: 87.5%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: overall. Source rank: #6. Votes: 49424. Organization: anthropic. License: Proprietary.

87.5% percentile inside its fair comparison set

1,487Raw benchmark valueCI 1,481 - 1,493

Vals Multimodal Index

VALS-AI · Document understanding · Combined

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

Rank #6 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 61
Percentile: 73.7%
Last updated: aging
Eligibility: headline eligible

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

73.7% percentile inside its fair comparison set

61Raw benchmark valueCI 58 - 63

MortgageTax

VALS-AI · Document understanding · Objective

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

Rank #18 · Source label: anthropic/claude-sonnet-4-6

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Vals AI
Raw value: 67.7%
Percentile: 71.7%
Last updated: aging
Eligibility: headline eligible

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

71.7% percentile inside its fair comparison set

67.7%Raw benchmark valueCI 65.9% - 69.5%

Multimodal mix

OC · Document understanding · Objective

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

Rank #12 · Source label: claude-sonnet-4

backfilledproxy backfilledBackground only

Raw row drilldownsource row, percentile, last updated, eligibility

Source: OpenCompass
Raw value: 72.7%
Percentile: 35.7%
Last updated: archived
Eligibility: Fallback benchmark identity is visible for context but excluded from default ranking.

Backfilled from Claude Sonnet 4 via approved benchmark identity mapping map-claude-sonnet-4-6-to-4.

35.7% percentile inside its fair comparison set

72.7%Raw benchmark value

Multilingual16 benchmarks88%

Text Arena · Chinese

AR · Multilingual · Human

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

Rank #25

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,503
Percentile: 91.9%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #30. Votes: 1919. Organization: anthropic. License: Proprietary.

91.9% percentile inside its fair comparison set

1,503Raw benchmark valueCI 1,489 - 1,518

Text Arena · French

AR · Multilingual · Human

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

Rank #27

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,478
Percentile: 88%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: french. Source rank: #35. Votes: 1159. Organization: anthropic. License: Proprietary.

88% percentile inside its fair comparison set

1,478Raw benchmark valueCI 1,459 - 1,498

Text Arena · German

AR · Multilingual · Human

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

Rank #21

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,460
Percentile: 91.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: german. Source rank: #27. Votes: 668. Organization: anthropic. License: Proprietary.

91.6% percentile inside its fair comparison set

1,460Raw benchmark valueCI 1,437 - 1,484

Text Arena · Japanese

AR · Multilingual · Human

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

Rank #26

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,423
Percentile: 87.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: japanese. Source rank: #36. Votes: 341. Organization: anthropic. License: Proprietary.

87.7% percentile inside its fair comparison set

1,423Raw benchmark valueCI 1,389 - 1,457

Text Arena · Korean

AR · Multilingual · Human

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

Rank #17

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,431
Percentile: 92.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: korean. Source rank: #21. Votes: 577. Organization: anthropic. License: Proprietary.

92.3% percentile inside its fair comparison set

1,431Raw benchmark valueCI 1,405 - 1,457

Text Arena · Russian

AR · Multilingual · Human

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

Rank #22

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,463
Percentile: 92.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: russian. Source rank: #31. Votes: 4306. Organization: anthropic. License: Proprietary.

92.7% percentile inside its fair comparison set

1,463Raw benchmark valueCI 1,453 - 1,472

Text Arena · Spanish

AR · Multilingual · Human

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

Rank #18

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,465
Percentile: 92.1%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: spanish. Source rank: #21. Votes: 1206. Organization: anthropic. License: Proprietary.

92.1% percentile inside its fair comparison set

1,465Raw benchmark valueCI 1,446 - 1,484

Text Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #31

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,483
Percentile: 89.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #36. Votes: 1919. Organization: anthropic. License: Proprietary.

89.8% percentile inside its fair comparison set

1,483Raw benchmark valueCI 1,469 - 1,498

Text Arena · French · No Style Control

AR · Multilingual · Human

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

Rank #32

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,461
Percentile: 85.6%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: french. Source rank: #38. Votes: 1159. Organization: anthropic. License: Proprietary.

85.6% percentile inside its fair comparison set

1,461Raw benchmark valueCI 1,442 - 1,480

Text Arena · German · No Style Control

AR · Multilingual · Human

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

Rank #37

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,432
Percentile: 84.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: german. Source rank: #46. Votes: 668. Organization: anthropic. License: Proprietary.

84.8% percentile inside its fair comparison set

1,432Raw benchmark valueCI 1,408 - 1,455

Text Arena · Japanese · No Style Control

AR · Multilingual · Human

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

Rank #32

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,402
Percentile: 84.7%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: japanese. Source rank: #42. Votes: 341. Organization: anthropic. License: Proprietary.

84.7% percentile inside its fair comparison set

1,402Raw benchmark valueCI 1,368 - 1,436

Text Arena · Korean · No Style Control

AR · Multilingual · Human

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

Rank #23

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,412
Percentile: 89.4%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: korean. Source rank: #29. Votes: 577. Organization: anthropic. License: Proprietary.

89.4% percentile inside its fair comparison set

1,412Raw benchmark valueCI 1,386 - 1,438

Text Arena · Russian · No Style Control

AR · Multilingual · Human

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

Rank #32

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,437
Percentile: 89.3%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: russian. Source rank: #40. Votes: 4306. Organization: anthropic. License: Proprietary.

89.3% percentile inside its fair comparison set

1,437Raw benchmark valueCI 1,427 - 1,447

Text Arena · Spanish · No Style Control

AR · Multilingual · Human

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

Rank #16

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,460
Percentile: 93%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: spanish. Source rank: #18. Votes: 1206. Organization: anthropic. License: Proprietary.

93% percentile inside its fair comparison set

1,460Raw benchmark valueCI 1,442 - 1,479

Vision Arena · Chinese

AR · Multilingual · Human

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

Rank #15

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,302
Percentile: 81.8%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #20. Votes: 997. Organization: anthropic. License: Proprietary.

81.8% percentile inside its fair comparison set

1,302Raw benchmark valueCI 1,277 - 1,326

Vision Arena · Chinese · No Style Control

AR · Multilingual · Human

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

Rank #21

verified runtimeexact alias

Raw row drilldownsource row, percentile, last updated, eligibility

Source: Arena
Raw value: 1,305
Percentile: 74%
Last updated: aging
Eligibility: headline eligible

Parsed from Arena leaderboard dataset row `claude-sonnet-4-6`. Category: chinese. Source rank: #26. Votes: 997. Organization: anthropic. License: Proprietary.

74% percentile inside its fair comparison set

1,305Raw benchmark valueCI 1,281 - 1,329