supermarket

Vision-language

Claude 3.5 Sonnet

By Anthropic, in the Claude Sonnet family.

Released
20 Jun 2024
Status
Retired
Retired
28 Oct 2025
Context
200K
Max output
8.2K
Input per 1M tokens
$3.00
Output per 1M tokens
$15.00
Open weights
No
License
proprietary
Parameters
—
Knowledge cutoff
Apr 2024
Reasoning
No
Modalities
text, image → text

Category scores

Out of 100, from independent and third-party results only, by method category-2026-09-29.1.

Category scores
CategoryScoreRankCoverage
Knowledge & reasoning29.7Not ranked1 of 4
Coding & software engineering28.1Not ranked2 of 5
Agentic tool use32.6Not ranked1 of 2
Professional knowledge work24.061 of 1
Vision & multimodal understanding31.0132 of 3
Human preference33.2Not ranked2 of 5

Benchmark results

Every result we hold for this model, with who ran it and where it was published.

Knowledge & reasoning

Knowledge & reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
DTBench (Decision Theory Benchmark)Accuracy (capability questions)67.8%Third party——https://conceptualreasoning.ai/dtbench via Epoch AI
DTBench (Decision Theory Benchmark)Accuracy (capability questions)67.8%Third party——https://conceptualreasoning.ai/dtbench via Epoch AI
MMLUAccuracy87.3%Third party——Stanford CRFM Leaderboard via Epoch AI
MMLUAccuracy86.5%Third party——Stanford CRFM Leaderboard via Epoch AI
SimpleBenchScore (AVG@5)41.4%Third party——SimpleBench Leaderboard via Epoch AI
SimpleBenchScore (AVG@5)27.5%Third party——SimpleBench Leaderboard via Epoch AI

Scientific reasoning

Scientific reasoning results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GPQA DiamondAccuracy55.3%Independent—27 Jan 2025Epoch AI
GPQA DiamondAccuracy54.0%Independent—27 Jan 2025Epoch AI

Mathematics

Mathematics results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
FrontierMath Tier 4 (v1)Accuracy0.0%Independent—1 Jul 2025Epoch AI
FrontierMath Tier 4 (v1)Accuracy0.0%Independent—1 Jul 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy2.1%Independent—6 Mar 2025Epoch AI
FrontierMath Tiers 1-3 (v1)Accuracy1.0%Independent—7 Mar 2025Epoch AI
MATH Level 5Accuracy56.9%Independent—27 Jan 2025Epoch AI
MATH Level 5Accuracy51.7%Independent—27 Jan 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy8.5%Independent—25 Feb 2025Epoch AI
OTIS Mock AIME 2024-2025Accuracy6.5%Independent—25 Feb 2025Epoch AI

Coding & software engineering

Coding & software engineering results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Aider PolyglotPercent correct51.6%Third party——https://aider.chat/docs/leaderboards/ via Epoch AI
CadEvalOverall pass rate48.0%Third party——CadEval Dashboard via Epoch AI
CybenchUnguided challenges solved17.5%Third party——Cybench leaderboard via Epoch AI
GSOOPT@14.6%Third party——GSO Leaderboard via Epoch AI
METR Time HorizonsAverage task success (Epoch's column)45.2%Third party——https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/ via Epoch AI
METR Time HorizonsAverage task success (Epoch's column)40.2%Third party——https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/ via Epoch AI
WeirdML (v2)Average accuracy40.0%Third party——https://htihle.github.io/weirdml.html via Epoch AI
WeirdML (v2)Average accuracy31.0%Third party——https://htihle.github.io/weirdml.html via Epoch AI

Agentic tool use

Agentic tool use results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
BALROGAverage progress32.6%Third party——https://balrogai.com/ via Epoch AI

Professional knowledge work

Professional knowledge work results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
TheAgentCompanyTasks resolved24.0%Third party——TheAgentCompany experiment results github via Epoch AI

Vision & multimodal understanding

Vision & multimodal understanding results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
GeoBenchCountry accuracy62.0%Third party——GeoBench leaderboard via Epoch AI
Visual Physics Comprehension Test (VPCT)Correct33.0%Third party——VPCT leaderboard via Epoch AI
Visual Physics Comprehension Test (VPCT)Correct33.0%Third party——VPCT leaderboard via Epoch AI

Writing

Writing results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
LLM Creative Story-Writing Benchmark V1 (Lech Mazur)Mean rubric score (percent of the 10-point scale)80.3%Third party——lechmazur/writing Github repository via Epoch AI

Human preference

Human preference results
BenchmarkMetricValueProvenanceSettingEvaluatedSource
Text Arena (arena.ai)Arena score (Elo scale)1374Independent—25 Sep 2026Arena
Text Arena (arena.ai)Arena score (Elo scale)1343Independent—25 Sep 2026Arena
Text Arena (arena.ai)Votes87644Independent—25 Sep 2026Arena
Text Arena (arena.ai)Votes82419Independent—25 Sep 2026Arena
Vision Arena (arena.ai)Arena score (Elo scale)1161Independent—28 Sep 2026Arena
Vision Arena (arena.ai)Arena score (Elo scale)1145Independent—28 Sep 2026Arena
Vision Arena (arena.ai)Votes21624Independent—28 Sep 2026Arena
Vision Arena (arena.ai)Votes10376Independent—28 Sep 2026Arena

Other Claude Sonnet versions

Sources