- Knowledge work GDPval-AA v2.1 1620
- Knowledge work AA-Briefcase v1.1 1578
- Computer use OSWorld 2.1 (Offline subset) 72.4
- Multidisciplinary reasoning Humanity's Last Exam (with tools) 57.4
- Agentic coding Terminal-Bench 4.0 39.2
- Agentic coding FrontierCode 1.1 (Main) (max) 46.4
- Visual reasoning Chartography (no tools) 46.4
- SWE-Bench Pro 64.8
- SWE-bench Multilingual 83.7
- SWE-bench Multimodal 30.7
- HealthBench Professional (Length-adjusted) 64.8
Claude Haiku 5.5
Claude Haiku 5.5 ranks first on Anthropic automated behavioral audit.
159
Benchmarks with a score
1
Ranked first, out of five or more models
7
In the top three
Anthropic Oct 2026
Claude Haiku 5.5
Closed weights Small
160.4
#14 of 271
Epoch Capabilities Index
- Humanity's Last Exam 30.1% #13 /47
- MultiChallenge 67.7% #10 /35
- HealthBench Professional 64.8% #6 /25
Where it ranks
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.
Knowledge and language
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Humanity's Last Exam | 30.1% | #13 of 47 | Official leaderboard | |
| BullshitBench V2 | 57.6% | #34 of 122 | Official leaderboard | |
| MultiChallenge | 67.7% | #10 of 35 | Official leaderboard | |
| LMArena (Elo) | 1,425 | #99 of 333 | Official leaderboard | |
| Extended NYT Connections | 65.7% | #56 of 97 | Official leaderboard | |
| Global MMLU Lite | 87.3% | #26 of 43 | Official leaderboard | |
| SimpleQA Verified | 23.9% | #37 of 56 | Official leaderboard | |
| FACTS Search V2 | 77.5% | #16 of 21 | Official leaderboard | |
| FACTS Grounding v2 | 57% | #12 of 15 | Official leaderboard | |
| Global MMLU | 87.8% | #9 of 11 | Model card | |
| FACTS Parametric | 13.1% | #37 of 45 | Official leaderboard | |
| MILU | 87.6% | #9 of 10 | Model card | |
| AA-Omniscience | 12% | #16 of 17 | Model card | |
| FACTS Parametric V2 | 13.1% | #20 of 20 | Official leaderboard |
Math and science
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.9% | #23 of 202 | Epoch AI | |
| ProofBench | 89% | #10 of 83 | Epoch AI | |
| DTBench | 91.7% | #23 of 178 | Official leaderboard | |
| CritPt | 22.6% | #25 of 160 | Epoch AI | |
| GPQA Diamond | 89.7% | #47 of 219 | Epoch AI | |
| ProofBench v1.1 | 89% | #11 of 51 | Official leaderboard | |
| FrontierMath Tiers 1-3 v2 | 75.1% | #21 of 89 | Epoch AI | |
| FrontierMath Tier 4 v2 | 46.3% | #24 of 71 | Epoch AI | |
| MysteryMechanism | 35.1% | #10 of 29 | Official leaderboard | |
| ARC-AGI-2 | 61.3% | #33 of 95 | Official leaderboard | |
| ARC-AGI | 89% | #36 of 92 | Official leaderboard | |
| FrontierPhysics | 4.20% | #10 of 19 | Official leaderboard | |
| Morphology-to-molecule matching (Anthropic) | 25.4% | #2 of 3 | Model card | |
| scBench (LatchBio SingleCellBench) | 56.2% | #5 of 6 | Model card | |
| SpatialBench (LatchBio) | 67.7% | #5 of 6 | Model card | |
| Terminal-Bench-Science | 20.6% | #9 of 9 | Model card | |
| Biomedical image analysis (Anthropic) | 43.5% | #3 of 3 | Model card | |
| Medicinal Chemistry (Anthropic) | 58% | #3 of 3 | Model card |
Coding
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Vibe Code Bench | 90.4% | #3 of 115 | Official leaderboard | |
| SWE-bench Multilingual | 83.7% | #9 of 70 | Model card | |
| ALE-Bench | 1,587 | #21 of 121 | Epoch AI | |
| MirrorCode | 78.9% | #5 of 28 | Epoch AI | |
| SWE-Bench Pro | 64.8% | #14 of 77 | Model card | |
| ProgramBench | 82% | #4 of 20 | Model card | |
| SWE-Marathon v1.1 | 50% | #4 of 17 | Official leaderboard | |
| Code Arena | 1,589 | #31 of 124 | Official leaderboard | |
| Brokk Power Ranking | 18.3 | #5 of 19 | Official leaderboard | |
| Code Migration | 44.6% | #21 of 78 | Official leaderboard | |
| CursorBench | 48.4% | #6 of 19 | Epoch AI | |
| FrontierCode | 46.4% | #15 of 46 | Epoch AI | |
| FrontierCode 2.0 | 27.7% | #9 of 26 | Official leaderboard | |
| FrontierSWE | 43.8% | #8 of 21 | Epoch AI | |
| Vibe Code Bench 1-100 | 20.4% | #11 of 26 | Official leaderboard | |
| Code Migration v1.1 | 52.4% | #13 of 29 | Official leaderboard | |
| OpenDesign Arena | 82.2 | #9 of 19 | Official leaderboard | |
| WeirdML v3 | 11.7% | #13 of 25 | Official leaderboard | |
| IOI v2 (Vals) | 47.3% | #30 of 45 | Official leaderboard | |
| FrontierCode 1.1 | 46.4% | #9 of 12 | Model card | |
| SWE-bench Multimodal | 30.7% | #9 of 10 | Model card | |
| SWE-Bench Pro V2 Hard | 52.9% | #15 of 16 | Official leaderboard | |
| SWE-Bench Pro V2 | 87.2% | #16 of 16 | Official leaderboard |
Agents and tools
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| LisanBench | 12897.0 | #11 of 133 | Official leaderboard | |
| LLM Chess | 1,143 | #17 of 124 | Official leaderboard | |
| Chess Puzzles | 55% | #14 of 83 | Epoch AI | |
| Game Arena Nine Men's Morris | 787 | #9 of 36 | Official leaderboard | |
| Game Arena Dots and Boxes | 966 | #12 of 37 | Official leaderboard | |
| Game Arena Go | 1,082 | #12 of 36 | Official leaderboard | |
| EBR-bench | 43.8% | #10 of 28 | Epoch AI | |
| Mystery Game Puzzles | 30% | #31 of 80 | Epoch AI | |
| Game Arena Reversi | 687 | #14 of 36 | Official leaderboard | |
| Game Arena Four in a Row | 416 | #16 of 38 | Official leaderboard | |
| Agent Arena | 0.01 | #23 of 53 | Official leaderboard | |
| ARC-AGI-3 | 2.82% | #11 of 24 | Official leaderboard | |
| Hermes Bench | 71% | #9 of 17 | Official leaderboard | |
| Game Arena Dark Hex | 239 | #22 of 38 | Official leaderboard | |
| Terminal-Bench 4.0 | 39.2% | #14 of 23 | Model card | |
| DRACO | 81.5% | #7 of 11 | Model card | |
| GAIA | 37.2% | #37 of 57 | Official leaderboard | |
| Kaggle Game Arena | 251 | #26 of 39 | Official leaderboard | |
| Game Arena Bargaining | 165 | #25 of 36 | Official leaderboard | |
| Game Arena Checkers | 643 | #28 of 38 | Official leaderboard | |
| Game Arena Chess | 594 | #29 of 39 | Official leaderboard | |
| Game Arena Coin Game | 232 | #28 of 35 | Official leaderboard | |
| Game Arena Clobber | 128 | #29 of 36 | Official leaderboard | |
| Game Arena Ultimate Tic-Tac-Toe | 85 | #29 of 36 | Official leaderboard | |
| StarSkirmish Bench | 21% | #8 of 9 | Official leaderboard | |
| Game Arena Word Art | 102 | #32 of 35 | Official leaderboard | |
| Game Arena Word Association | 20 | #33 of 36 | Official leaderboard | |
| Game Arena Lines of Action | 177 | #35 of 36 | Official leaderboard | |
| o11y-bench | 87.5% | #6 of 6 | Official leaderboard | |
| WANDR | 49.9% | #3 of 3 | Model card | |
| OSWorld 2.1 (offline subset) | 72.4% | #2 of 2 | Model card |
Professional work
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| MedScribe | 89.5% | #6 of 98 | Official leaderboard | |
| MedCode | 51.5% | #15 of 96 | Official leaderboard | |
| EEBench | 67.1% | #5 of 26 | Official leaderboard | |
| HealthBench Professional | 64.8% | #6 of 25 | Model card | |
| Vals Finance Agent v2.1 | 65.4% | #9 of 35 | Official leaderboard | |
| HealthBench | 61.6% | #8 of 30 | Model card | |
| Legal Research Bench | 43.8% | #25 of 88 | Official leaderboard | |
| Tax Agent Bench | 31.9% | #23 of 79 | Official leaderboard | |
| Vals Finance Agent v2 | 54.1% | #28 of 80 | Official leaderboard | |
| APEX-Agents | 56.1% | #20 of 56 | Epoch AI | |
| Vals Legal Index | 70.3% | #14 of 33 | Official leaderboard | |
| Vals Finance Index | 57.5% | #15 of 35 | Official leaderboard | |
| Vals Index v2.2 | 58.7% | #14 of 32 | Official leaderboard | |
| Excel Modeling Benchmark | 59% | #37 of 76 | Official leaderboard | |
| Public Benefits Bench | 63.2% | #25 of 50 | Official leaderboard | |
| RCM-Bench-1 | 88.4% | #10 of 20 | Official leaderboard | |
| OfficeQA | 73.5% | #7 of 12 | Model card | |
| Vending-Bench 2 | $3,602 | #45 of 66 | Official leaderboard | |
| TaxCalcBench v2 | 22% | #18 of 26 | Official leaderboard | |
| BenchCAD | 67% | #7 of 10 | Model card | |
| Harvey LAB (Legal Agent Benchmark) | 1.25% | #58 of 80 | Official leaderboard | |
| GDPval-AA v2.1 | 1,620 | #6 of 8 | Model card | |
| Excel Modeling Benchmark v1.1 | 65.5% | #29 of 37 | Official leaderboard | |
| PolicyBench | 79.2% | #40 of 50 | Official leaderboard | |
| AA-Briefcase v1.1 | 1,578 | #5 of 6 | Model card | |
| OfficeQA Pro | 60.3% | #12 of 14 | Model card | |
| Immigration Bench | 61.8% | #21 of 24 | Official leaderboard | |
| Harvey LAB v1.1 (HLAB-Vals-1.1) | 4.17% | #29 of 33 | Official leaderboard | |
| PhysicianBench | 43% | #2 of 2 | Model card |
Vision, audio and robotics
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| Furniture Assembly Benchmark | 47.5% | #13 of 35 | Epoch AI | |
| Blueprint-Bench 2 | 33% | #15 of 35 | Official leaderboard | |
| VoxelBench | 1,757 | #27 of 56 | Official leaderboard | |
| 1H-VideoQA | 68.3% | #13 of 20 | Official leaderboard | |
| ParseBench | 56.1% | #29 of 42 | Official leaderboard | |
| Chartography | 46.4% | #10 of 12 | Model card |
Safety and security
Other
| Benchmark | Score | Rank | Source | |
|---|---|---|---|---|
| ACCoRD | 81.5 | #7 of 150 | Official leaderboard | |
| SAGE v1.1 (Vals) | 89.8% | #3 of 28 | Official leaderboard | |
| Conceptual Reasoning Index | 65.3 | #21 of 149 | Official leaderboard | |
| LMCA | 51.1% | #24 of 139 | Epoch AI | |
| LiveBench | 69.8% | #12 of 28 | Official leaderboard | |
| SAGE (Vals) | 45.8% | #38 of 81 | Official leaderboard | |
| Anthropic ECI (AECI) | 167.1 | #3 of 6 | Model card | |
| Vals Science Index | 38.5% | #12 of 23 | Official leaderboard | |
| Hermes Index | 37.0 | #10 of 17 | Official leaderboard |
Anthropic's models over time
Anthropic's models in color, against every other model on the Epoch Capabilities Index, by release date.
Its launch post
The benchmarks the lab chose to headline, with the scores it gave.
Its system card
Every result in the longer documents the lab published with it. 58 of them aren't in the launch post.
System Card: Claude Haiku 5.5 7 Oct 2026
Capabilities
- SWE-Bench Pro 64.8
- SWE-bench Multilingual 83.7
- SWE-bench Multimodal 30.7
- FrontierCode 1.1 46.4
- Humanity's Last Exam 45.9
- OSWorld 2.1 (offline subset) 72.4
- GDPval-AA v2.1 1620
- AA-Briefcase v1.1 1578
- Terminal-Bench 4.0 39.2
- Terminal-Bench-Science 20.6
- FrontierSWE 43.8
- ProgramBench 82
- DRACO 81.5
- WANDR 49.9
- Chartography 46.4
- BenchCAD 67
- OfficeQA 73.5
- OfficeQA Pro 60.3
- Global MMLU 87.8
- MILU 87.6
- Morphology-to-molecule matching (Anthropic) 25.4
- Medicinal Chemistry (Anthropic) 58
- Protein Design (Anthropic)
- Biomedical image analysis (Anthropic) 43.5
Health
- HealthBench Professional 64.8
- HealthBench 61.6
- PhysicianBench 43
Hallucination
Safety
- Single-turn harmful request evaluations (Anthropic) 98.4
- Single-turn benign request evaluations (Anthropic) 0.17
- Multi-turn testing (Anthropic)
- Child safety evaluations (Anthropic)
- Suicide and self-harm evaluations (Anthropic)
- Disordered eating evaluations (Anthropic)
- Election integrity evaluations (Anthropic)
- Malicious use of Claude Code (Anthropic) 84.3
- Malicious computer use (Anthropic) 82.6
Robustness
Alignment
- Anthropic automated behavioral audit 3.19
- Anthropic undermining liberal democracy (behavioral audit metric) 1.22
- Sandbox escape evaluation (Anthropic) 8.7
- Opportunistic sandbox escape (Anthropic) 4
- Self-preference bias (Anthropic) 0.23
- MASK 96.8
- Silent copying of answers (Anthropic) 17.3
- Concealment of changes (Anthropic) 95.9
Monitorability
Bio and chem risk
Cyber risk
AI R&D risk
- Anthropic ECI (AECI) 167.1