Model
Claude Sonnet 4.6
Anthropic · closed weights
Knowledge & Reasoning
Knowledge & Reasoning| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| CritPt | 3.1 | accuracy | N/A | Independent | low | epoch.ai |
| CroissantMiner | 0.7 | accuracy | single-pass | Independent | medium | arxiv.org |
| GPQA Diamond | 89.9 | accuracy | N/A | Self-reported | low | artificialanalysis.ai |
| GPQA Diamond | 83.3 | accuracy | N/A | Independent | high | epoch.ai |
| HLE | 49.0 | accuracy | N/A | Self-reported | low | lastexam.ai |
| LMArena | 1460.0 | elo | N/A | Self-reported | low | lmarena.ai |
| MMLU-Pro | 79.1 | accuracy | N/A | Self-reported | low | artificialanalysis.ai |
| SimpleQA Verified | 35.5 | accuracy | N/A | Independent | high | epoch.ai |
| VERITYGATE | 52.1 | claim_survival_rate | veritygate | Independent | medium | arxiv.org |
Math
Math| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| AIME 2025 | 52.8 | accuracy | N/A | Self-reported | low | kaggle.com |
| MATH-500 | 97.8 | accuracy | N/A | Self-reported | low | kaggle.com |
Coding
Coding| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| DeepSWE | 29.9 | pass@1 | mini-swe-agent | Independent | medium | deepswe.datacurve.ai |
| LiveCodeBench | 72.4 | pass@1 | N/A | Self-reported | low | livecodebench.github.io |
| MathVis | 16.8 | pass@1 | N/A | Independent | medium | arxiv.org |
| ParanoiaEval | 11.2 | violation_rate | N/A | Independent | medium | arxiv.org |
| SciCode | 46.8 | score | N/A | Independent | medium | artificialanalysis.ai |
| SWE-bench Verified | 79.6 | resolved_rate | N/A | Self-reported | low | swebench.com |
| SWE-bench Verified | 75.2 | resolved_rate | N/A | Independent | high | epoch.ai |
| Terminal-Bench 2.0 | 59.1 | accuracy | N/A | Self-reported | low | tbench.ai |
Agentic / Tool Use
Agentic / Tool Use| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| BrowseComp | 74.7 | accuracy | N/A | Self-reported | low | openai.com |
| DeepResearch Bench | 54.9 | average_score | N/A | Independent | medium | evals.futuresearch.ai |
| ExploitBench | 23.6 | mean_capability | N/A | Independent | low | epoch.ai |
| FinSkillBench | 0.6 | score | N/A | Independent | medium | arxiv.org |
| Legal Research Bench | 38.5 | accuracy | N/A | Independent | medium | arxiv.org |
| MisVisAgentBench | 34.3 | pra | N/A | Independent | medium | arxiv.org |
| OSWorld | 72.5 | success_rate | N/A | Self-reported | low | os-world.github.io |
| OSWorld | 72.1 | score | claude-sonnet-4-6 (100 steps) | Independent | medium | os-world.github.io |
| τ²-bench | 91.7 | pass^1 | N/A | Self-reported | low | github.com |
| Terminal-Bench | 53.4 | accuracy | Simplai Agent | Independent | medium | tbench.ai |
| UserProxyBench | 0.9 | ufs | N/A | Independent | medium | arxiv.org |
Multimodal
Multimodal| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| DataVista | 63.6 | accuracy | N/A | Independent | medium | arxiv.org |
| GDP.pdf | 18.0 | gdp.pdf_score | N/A | Independent | medium | surgehq.ai |
| MMMU-Pro | 75.6 | accuracy | N/A | Self-reported | low | artificialanalysis.ai |
| R-GroundBench | 70.5 | accuracy | N/A | Independent | medium | arxiv.org |
| SpaceCast-Bench | 36.0 | accuracy | N/A | Independent | medium | arxiv.org |
| SpatialChain | 79.9 | accuracy | N/A | Independent | medium | arxiv.org |
Instruction Following
Instruction FollowingSafety & Alignment
Safety & Alignment| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| MediateRec | 13.5 | resc. | N/A | Independent | medium | arxiv.org |
| Mental Health Stigma Benchmark | 0.6 | f1 | N/A | Independent | medium | arxiv.org |
Abstract Reasoning
Abstract Reasoning| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| ARC-AGI-2 | 60.4 | score | N/A | Independent | low | epoch.ai |
| ARC-AGI-2 | 58.3 | accuracy | N/A | Self-reported | low | arcprize.org |
Reproduced by the community
Loading community runs…