Claude Sonnet 4 ranks first on 8 of the 205 benchmarks with five or more models scored.
Closed weights
Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Meeseeks Utility Rate | 90.5% | #1 of 10 | Official leaderboard | |
| MMLU-Pro | 83.7% | #38 of 212 | Official leaderboard | |
| AraGen | 75.6% | #8 of 42 | Official leaderboard | |
| LLM Confabulation Leaderboard | 13.2% | #13 of 58 | Official leaderboard | |
| Slop Score | 27.3 | #4 of 17 | Official leaderboard | |
| Video SimpleQA | 35.6% | #7 of 27 | Official leaderboard | |
| Judgemark v2.1 | 82.0 | #11 of 42 | Official leaderboard | |
| SuperGPQA | 57% | #17 of 59 | Official leaderboard | |
| MMMLU | 86.5% | #23 of 71 | Lab model card | |
| FACTS Grounding | 79.1% | #13 of 34 | Official leaderboard | |
| LMArena (Elo) | 1,402 | #129 of 335 | Official leaderboard | |
| IFEval | 84% | #24 of 60 | Official leaderboard | |
| SnorkelSpatial | 33.3% | #11 of 27 | Official leaderboard | |
| WritingBench | 74.1 | #19 of 45 | Official leaderboard | |
| RewardBench 2 | 71.2% | #18 of 42 | Official leaderboard | |
| EQ-Bench 3 | 1,304 | #35 of 78 | Official leaderboard | |
| AMemGym | 46.2% | #5 of 11 | Official leaderboard | |
| ECLeKTic | 18.5% | #17 of 35 | Official leaderboard | |
| Global MMLU Lite | 90.6% | #21 of 43 | Official leaderboard | |
| MultiChallenge | 57.1% | #21 of 42 | Official leaderboard | |
| WideSearch | 3.60% | #3 of 6 | Official leaderboard | |
| FaithJudge | 18.2% | #19 of 37 | Official leaderboard | |
| BullshitBench | 25.4% | #56 of 107 | Official leaderboard | |
| Vectara Hallucination Leaderboard (HHEM) | 4.50% | #79 of 148 | Official leaderboard | |
| Longform Creative Writing (EQ-Bench) | 58.4 | #68 of 119 | Official leaderboard | |
| Creative Writing v3 | 65.4 | #70 of 120 | Official leaderboard | |
| Vectara Hallucination Leaderboard (2025 dataset) | 10.3% | #61 of 104 | Official leaderboard | |
| BullshitBench V2 | 30% | #78 of 129 | Official leaderboard | |
| FACTS Search | 66.3% | #19 of 31 | Official leaderboard | |
| HUMAINE | 23 | #34 of 50 | Official leaderboard | |
| FACTS Parametric | 20.4% | #34 of 45 | Official leaderboard | |
| Humanity's Last Exam | 7.76% | #36 of 47 | Official leaderboard | |
| Fiction.LiveBench | 46.9% | #45 of 57 | Fiction.live leaderboard, via Epoch AI | |
| MultiNRC | 18.4% | #30 of 38 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Misguided Attention | 68.6% | #2 of 45 | Official leaderboard | |
| GPQA Main | 70.6% | #10 of 60 | Official leaderboard | |
| CORE-Bench | 46.7% | #4 of 17 | Official leaderboard | |
| SnorkelSequences | 70.4% | #6 of 25 | Official leaderboard | |
| MATH Level 5 | 84.4% | #23 of 94 | Epoch AI run | |
| SciArena | 1,025 | #13 of 34 | Official leaderboard | |
| HiPhO | 58% | #14 of 36 | Official leaderboard | |
| ForecastBench | 0.13 | #32 of 82 | Official leaderboard | |
| MATH-500 | 93.8% | #15 of 38 | Official leaderboard | |
| GPQA Diamond | 79.2% | #99 of 229 | Epoch AI run | |
| OTIS Mock AIME 2024-2025 | 71.1% | #106 of 212 | Epoch AI run | |
| AIME 2025 | 85% | #73 of 143 | Lab model card | |
| SnorkelGraph | 58% | #13 of 24 | Official leaderboard | |
| DTBench | 77.1% | #99 of 180 | Official leaderboard | |
| IneqMath | 3.00% | #20 of 34 | Official leaderboard | |
| SimpleBench | 45.5% | #68 of 106 | Official leaderboard | |
| PHYBench | 19.2% | #18 of 28 | Official leaderboard | |
| ARC-AGI-2 | 5.93% | #65 of 98 | Official leaderboard | |
| AstaBench | 37.5% | #9 of 13 | Official leaderboard | |
| ARC-AGI | 40% | #72 of 95 | Official leaderboard | |
| FrontierMath | 4.14% | #44 of 55 | Official leaderboard | |
| IMO-ProofBench | 4.80% | #17 of 20 | Official leaderboard | |
| FrontierMath Tier 4 | 0% | #50 of 55 | Epoch AI run | |
| CMT-Benchmark | 6.00% | #10 of 11 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| SWE-bench Verified | 76.8% | #3 of 60 | Official leaderboard | |
| SArena | 0.92 | #1 of 17 | Official leaderboard | |
| SWE-bench Lite | 60.3% | #1 of 15 | Official leaderboard | |
| AutoCodeBench | 51.1% | #3 of 34 | Official leaderboard | |
| ArtifactsBench | 57.3% | #4 of 20 | Official leaderboard | |
| CATArena SWE-Perf-10 | 1.95× | #1 of 5 | Official leaderboard | |
| SWE-bench Multimodal | 35.6% | #2 of 9 | Official leaderboard | |
| CC-Bench | 41.9% | #1 of 4 | Official leaderboard | |
| Aider Polyglot | 61.3% | #13 of 46 | Official leaderboard | |
| METR Time Horizon | 74 min | #9 of 31 | Official leaderboard | |
| SWE-Bench Pro | 42.7% | #9 of 24 | Official leaderboard | |
| CATArena OIBench-5 | 0.31 | #2 of 5 | Official leaderboard | |
| CATArena Texas Holdem | 1,372 | #2 of 5 | Official leaderboard | |
| WeirdML | 46.1% | #56 of 112 | Official leaderboard | |
| CodeClash | 1,224 | #4 of 8 | Official leaderboard | |
| ts-bench v1 | 92% | #5 of 9 | Official leaderboard | |
| DesignArena 3D Design | 1,122 | #83 of 136 | Official leaderboard | |
| DesignArena Game Dev | 1,129 | #92 of 143 | Official leaderboard | |
| Time series forecasting task (Anthropic) | 5.80 | #6 of 9 | Lab model card | |
| BIRD-Interact | 23.3% | #2 of 3 | Official leaderboard | |
| METR data deduplication | 27.6% | #2 of 3 | Lab model card | |
| DesignArena Website | 1,139 | #104 of 151 | Official leaderboard | |
| IOI (Vals) | 6.50% | #38 of 55 | Official leaderboard | |
| GSO-Bench | 4.90% | #21 of 30 | Official leaderboard | |
| DesignArena Code | 1,132 | #102 of 143 | Official leaderboard | |
| LiveCodeBench | 74.7% | #15 of 21 | Official leaderboard | |
| LiveCodeBench v5 | 57.9% | #15 of 21 | Official leaderboard | |
| LiveCodeBench v6 | 55.9% | #15 of 21 | Official leaderboard | |
| Text-based reinforcement learning task (Anthropic) | 0.68 | #5 of 7 | Lab model card | |
| DesignArena UI Component | 1,120 | #100 of 139 | Official leaderboard | |
| LiveSQLBench | 20% | #26 of 36 | Official leaderboard | |
| ALE-Bench | 1,088 | #96 of 126 | Official leaderboard | |
| DesignArena Data Visualization | 1,155 | #90 of 118 | Official leaderboard | |
| DesignArena SVG | 1,069 | #79 of 103 | Official leaderboard | |
| SWE-bench Verified (hard subset) | 36.7% | #7 of 9 | Lab model card | |
| CATArena Gomoku | 868 | #4 of 5 | Official leaderboard | |
| Internal AI research evaluation suite 2 | 0.36 | #5 of 6 | Lab model card | |
| DAXBench v2 | 58.4% | #47 of 47 | Official leaderboard | |
| SWE-Bench Pro Private | 9.06% | #12 of 12 | Official leaderboard | |
| Quadruped reinforcement learning task (Anthropic) | 0.35 | #8 of 8 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| C3-Bench | 12.5% | #1 of 30 | Official leaderboard | |
| tau-bench | 70.3% | #2 of 23 | Lab model card | |
| CompileBench | 93.3% | #2 of 20 | Official leaderboard | |
| WARC-Bench | 64.8% | #1 of 8 | Official leaderboard | |
| SnakeBench | 32 | #27 of 188 | Official leaderboard | |
| Meta Agents Research Environments (ARE) / Gaia2 | 37.8% | #2 of 13 | Official leaderboard | |
| AgentBench FC | 58.2% | #2 of 12 | Official leaderboard | |
| Galileo Agent Leaderboard v2 | 55% | #4 of 21 | Official leaderboard | |
| MCP-Universe | 32.9% | #6 of 30 | Official leaderboard | |
| MCP-Bench | 0.68 | #4 of 15 | Official leaderboard | |
| PinchBench | 74% | #14 of 48 | Official leaderboard | |
| LisanBench | 2236.0 | #42 of 133 | Official leaderboard | |
| Elimination Game | 5 | #18 of 56 | Official leaderboard | |
| OSWorld-MCP | 45% | #3 of 9 | Official leaderboard | |
| MCPMark | 28.4% | #10 of 28 | Official leaderboard | |
| Terminal-Bench | 41.3% | #8 of 22 | Lab model card | |
| Step Race | 3 | #23 of 59 | Official leaderboard | |
| TextQuests | 16.0 | #11 of 28 | Official leaderboard | |
| lmgame-Bench | 31.1 | #10 of 23 | Official leaderboard | |
| GAIA | 55.5% | #31 of 64 | Official leaderboard | |
| LLM Chess | 115 | #60 of 123 | Official leaderboard | |
| OSWorld | 43.9% | #4 of 8 | OS World Website, via Epoch AI | |
| BrowseComp-Plus | 36.8% | #4 of 7 | Official leaderboard | |
| SnorkelWordle | 83% | #11 of 19 | Official leaderboard | |
| DeepResearch Bench | 46.6% | #13 of 22 | evals.futuresearch.ai, via Epoch AI | |
| OSWorld-Verified | 43.9% | #20 of 33 | Official leaderboard | |
| Deep Research Bench (FutureSearch) | 0.47 | #13 of 21 | Official leaderboard | |
| Toolathlon | 29.9% | #31 of 50 | Official leaderboard | |
| Research-Eval | 44.8% | #2 of 3 | Official leaderboard | |
| LiveResearchBench | 62.3 | #5 of 7 | Official leaderboard | |
| RuneBench | 3.71 | #62 of 76 | Official leaderboard | |
| PinchBench 2.0 | 48.8% | #49 of 57 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| TheAgentCompany | 33.1% | #1 of 7 | Official leaderboard | |
| Vals Finance Agent v1 | 44.5% | #10 of 47 | Official leaderboard | |
| Finance Reasoning (Snorkel) | 49.4% | #5 of 21 | Official leaderboard | |
| SnorkelUnderwrite | 72.3% | #11 of 27 | Official leaderboard | |
| FinQA (Snorkel OpenEnv environment) | 76.6% | #9 of 22 | Official leaderboard | |
| SnorkelFinance | 76.6% | #9 of 22 | Official leaderboard | |
| LegalBench | 83% | #63 of 138 | Official leaderboard | |
| CorpFin v2 | 61.2% | #56 of 119 | Official leaderboard | |
| TaxEval v2 | 72% | #66 of 129 | Official leaderboard | |
| StockBench | 7.00 | #7 of 13 | Official leaderboard | |
| BizFinBench v2 | 39.5% | #5 of 9 | Official leaderboard | |
| MortgageTax | 62.5% | #51 of 87 | Official leaderboard | |
| MedCode | 35% | #73 of 96 | Official leaderboard | |
| ProfBench | 42.5% | #55 of 72 | Official leaderboard | |
| MedScribe | 72.4% | #79 of 97 | Official leaderboard | |
| PLawBench | 53.5% | #9 of 11 | Official leaderboard | |
| TaxCalcBench | 23% | #16 of 17 | Official leaderboard | |
| OccuBench | 53.4% | #14 of 14 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MMMU | 74.4% | #24 of 136 | Official leaderboard | |
| VisualPuzzles | 47.3% | #7 of 29 | Official leaderboard | |
| MathCanvas-Bench | 25% | #5 of 16 | Official leaderboard | |
| MME-Reasoning | 36.1% | #4 of 11 | Official leaderboard | |
| VISTA | 45.5% | #25 of 54 | Official leaderboard | |
| Vision Arena | 1,206 | #76 of 136 | Official leaderboard | |
| ChartMuseum | 52.6% | #12 of 21 | Official leaderboard | |
| Math-VR | 28.1% | #15 of 26 | Official leaderboard | |
| ZeroBench | 6.00% | #48 of 76 | Official leaderboard | |
| MeasureBench Real-world | 12.6% | #12 of 19 | Official leaderboard | |
| MeasureBench Synthetic | 11% | #12 of 19 | Official leaderboard | |
| Blueprint-Bench | 0.27 | #7 of 8 | Official leaderboard | |
| VisFactor | 17.4% | #23 of 26 | Official leaderboard | |
| VPCT | 34% | #24 of 27 | VPCT leaderboard, via Epoch AI | |
| ClockBench | 6.70% | #34 of 37 | Official leaderboard | |
| GeoBench | 37% | #47 of 48 | Official leaderboard |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| AIR-Bench 2024 | 88.3% | #4 of 82 | Official leaderboard | |
| HELM Safety | 98.1 | #4 of 75 | Official leaderboard | |
| BaxBench | 56.1% | #2 of 34 | Official leaderboard | |
| AutoBaxBench | 37.9% | #2 of 11 | Official leaderboard | |
| nocot-bench (NCRI) | 91.3 | #35 of 185 | Official leaderboard | |
| SimpleSafetyTests | 100% | #15 of 75 | Official leaderboard | |
| Lakera B3 (Backbone Breaker Benchmark) | 74.7% | #7 of 35 | Official leaderboard | |
| Average in-eval reward hacking rate (Anthropic) | 14.3% | #1 of 5 | Lab model card | |
| ABC Bench Fragment Design | 96% | #6 of 27 | Official leaderboard | |
| Single-turn harmful request evaluations (Anthropic) | 99% | #6 of 19 | Lab model card | |
| FORTRESS | 18.1% | #26 of 73 | Official leaderboard | |
| Malicious use of Claude Code (Anthropic) | 90% | #7 of 17 | Lab model card | |
| Anthropic Red Team | 99.2% | #31 of 75 | Official leaderboard | |
| BBQ (Bias Benchmark for QA) | 86.3% | #11 of 25 | Lab model card | |
| Impossible tasks (Anthropic) | 45% | #5 of 11 | Lab model card | |
| Cybench | 56.4% | #9 of 18 | Lab model card | |
| HelpBench | 82% | #5 of 9 | Official leaderboard | |
| Molecular Biology Capabilities Test | 46.4% | #54 of 90 | Official leaderboard | |
| Reward-hack-prone coding tasks (Anthropic) | 4.00% | #5 of 8 | Lab model card | |
| Spiral-Bench v1.2 | 32.1 | #14 of 21 | Official leaderboard | |
| Single-turn benign request evaluations (Anthropic) | 0.23% | #13 of 19 | Lab model card | |
| SusVibes | 11.8% | #7 of 10 | Official leaderboard | |
| SCONE-bench | $331,500,000 | #5 of 7 | Official leaderboard | |
| Human Pathogen Capabilities Test | 42% | #64 of 86 | Official leaderboard | |
| BinaryAudit | 14% | #15 of 20 | Official leaderboard | |
| Prompt injection robustness against adaptive attackers, browser use (Anthropic) | 2.00% | #3 of 4 | Official leaderboard | |
| CyberGym | 22.6% | #26 of 34 | Official leaderboard | |
| ABC Bench Liquid-Handling Robot | 51.3% | #28 of 34 | Official leaderboard | |
| Network CTF challenges (Anthropic) | 25% | #5 of 6 | Lab model card | |
| Rev CTF challenges (Anthropic) | 37.5% | #5 of 6 | Lab model card | |
| World-Class Bio | 27% | #65 of 74 | Official leaderboard | |
| Spiral-Bench v1.0 | 41.1 | #12 of 13 | Official leaderboard | |
| ABC Bench Screening Evasion | 39.2% | #16 of 17 | Official leaderboard | |
| Creative biology | 37% | #7 of 7 | Lab model card | |
| Crypto CTF challenges (Anthropic) | 31.8% | #6 of 6 | Lab model card | |
| Pwn CTF challenges (Anthropic) | 33.3% | #6 of 6 | Lab model card | |
| Web CTF challenges (Anthropic) | 73.3% | #6 of 6 | Lab model card | |
| Computer use prompt injection evaluation (Anthropic) | 69% | #5 of 5 | Lab model card | |
| Bioweapons acquisition uplift trial (Anthropic) | 1.70 | #3 of 3 | Lab model card | |
| Bioweapons knowledge questions | 51.5% | #3 of 3 | Lab model card |
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| HELM Capabilities | 76.6 | #11 of 60 | Official leaderboard | |
| CAIS Risk Index | 47.1 | #27 of 67 | Official leaderboard | |
| Conceptual Reasoning Index | 41.1 | #98 of 151 | Official leaderboard | |
| LMCA | 29% | #96 of 147 | Official leaderboard | |
| ACCoRD | 57.0 | #100 of 152 | Official leaderboard | |
| SAGE (Vals) | 35% | #67 of 81 | Official leaderboard | |
| CAIS Text Capabilities Index | 17.1 | #57 of 68 | Official leaderboard | |
| CAIS Vision Capabilities Index | 26.4 | #53 of 53 | Official leaderboard |
The Self-Reported Index fits one score to the 8 capability results Anthropic reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 140.5 to 147.2. How we compute it.
Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.
| Benchmark | Lab says | Board says | Gap |
|---|---|---|---|
| GPQA Diamond | 83.8% Graduate-level reasoning GPQA Diamond (parallel test-time compute) | 79.2% Epoch AI run, 59K | +4.6 pts |
| SWE-bench Verified | 80.2% Agentic coding SWE-bench Verified (parallel test-time compute) | 76.8% Leaderboard, EPAM AI/Run Developer Agent | +3.4 pts |
| MMMU | 74.4% Visual reasoning MMMU (validation) | 74.4% Leaderboard | Same |
Anthropic's models in color, against every other model on the Epoch Capabilities Index, by release date.
Source: Epoch Capabilities Index, Epoch AI (CC BY 4.0)
The benchmarks the lab chose to headline, with the scores it gave.
Every result in the longer documents the lab published with it. 42 of them aren't in the launch post.