Model
Claude Opus 4.6
Anthropic · closed weights
Knowledge & Reasoning
Knowledge & Reasoning| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| CL-bench | 20.7 | overall | N/A | Independent | low | epoch.ai |
| CL-bench Life | 17.0 | overall | N/A | Independent | low | epoch.ai |
| EnigmaEval | 7.6 | accuracy | N/A | Independent | medium | labs.scale.com |
| GPQA Diamond | 91.3 | accuracy | N/A | Self-reported | low | artificialanalysis.ai |
| GPQA Diamond | 88.4 | accuracy | N/A | Independent | high | epoch.ai |
| HLE | 53.0 | accuracy | N/A | Self-reported | low | lastexam.ai |
| HLE | 34.4 | accuracy | N/A | Independent | low | epoch.ai |
| LMArena | 1503.0 | elo | N/A | Self-reported | low | lmarena.ai |
| MMLU-Pro | 82.0 | accuracy | N/A | Self-reported | low | artificialanalysis.ai |
| SimpleQA Verified | 47.0 | accuracy | N/A | Independent | high | epoch.ai |
Math
Math| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| AIME 2025 | 100.0 | accuracy | N/A | Self-reported | low | kaggle.com |
| EfficientOpt | 73.0 | solver_time_ratio | N/A | Independent | medium | arxiv.org |
| MATH-500 | 97.6 | accuracy | N/A | Self-reported | low | kaggle.com |
Coding
Coding| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| CRBugBench | 45.4 | success_rate | N/A | Independent | medium | arxiv.org |
| GSO | 41.2 | score_opt@1 | OpenHands | Independent | medium | epoch.ai |
| LiveCodeBench | 76.0 | pass@1 | N/A | Self-reported | low | livecodebench.github.io |
| QTestBench | 0.9 | test_quality | N/A | Independent | medium | arxiv.org |
| SWE-bench Verified | 80.8 | resolved_rate | N/A | Self-reported | low | swebench.com |
| SWE-bench Verified | 78.7 | resolved_rate | N/A | Independent | high | epoch.ai |
| Terminal-Bench 2.0 | 65.4 | accuracy | N/A | Self-reported | low | tbench.ai |
Agentic / Tool Use
Agentic / Tool Use| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| BrowseComp | 84.0 | accuracy | N/A | Self-reported | low | openai.com |
| DeepResearch Bench | 55.3 | average_score | N/A | Independent | medium | evals.futuresearch.ai |
| MisVisAgentBench | 41.4 | pra | N/A | Independent | medium | arxiv.org |
| OSWorld | 72.7 | success_rate | N/A | Self-reported | low | os-world.github.io |
| τ²-bench | 91.9 | pass^1 | N/A | Self-reported | low | github.com |
| Terminal-Bench | 79.8 | accuracy | ForgeCode | Independent | medium | tbench.ai |
| Terminal-Bench | 76.4 | accuracy | Meta-Harness | Independent | medium | tbench.ai |
| Terminal-Bench | 75.3 | accuracy | Capy | Independent | medium | tbench.ai |
| Terminal-Bench | 74.7 | accuracy | Terminus-KIRA | Independent | medium | tbench.ai |
| Terminal-Bench | 72.1 | accuracy | MAYA-V2 | Independent | medium | tbench.ai |
| Terminal-Bench | 71.9 | accuracy | TongAgents | Independent | medium | tbench.ai |
| Terminal-Bench | 69.9 | accuracy | Droid | Independent | medium | tbench.ai |
| Terminal-Bench | 66.9 | accuracy | Crux | Independent | medium | tbench.ai |
| Terminal-Bench | 66.5 | accuracy | Mux | Independent | medium | tbench.ai |
| Terminal-Bench | 62.9 | accuracy | Terminus 2 | Independent | medium | tbench.ai |
| Terminal-Bench | 58.0 | accuracy | Claude Code | Independent | medium | tbench.ai |
Multimodal
Multimodal| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| BrainTRACE | 29.8 | accuracy | N/A | Independent | medium | arxiv.org |
| ChartBmkAgent | 61.3 | accuracy | N/A | Independent | medium | arxiv.org |
| EgoExo-Next | 33.6 | accuracy | N/A | Independent | medium | arxiv.org |
| MMMU-Pro | 77.3 | accuracy | N/A | Self-reported | low | artificialanalysis.ai |
| NextAct | 0.3 | embedding_score | N/A | Independent | medium | arxiv.org |
| OccludeBench | 63.0 | accuracy | N/A | Independent | medium | arxiv.org |
Instruction Following
Instruction Following| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| IFEval | 94.0 | prompt_strict_accuracy | N/A | Self-reported | low | artificialanalysis.ai |
Safety & Alignment
Safety & Alignment| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| Cybench | 93.0 | unguided_%_solved | N/A | Self-reported | medium | www-cdn.anthropic.com |
| MediateRec | 13.6 | resc. | N/A | Independent | medium | arxiv.org |
| RAGStress | 99.7 | accuracy | N/A | Independent | medium | arxiv.org |
Abstract Reasoning
Abstract Reasoning| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| ARC-AGI-2 | 68.8 | accuracy | N/A | Self-reported | low | arcprize.org |
Reproduced by the community
Loading community runs…