Claude Sonnet 4.6 (High): benchmark results
Claude Sonnet 4.6 evaluated at the high reasoning-effort setting. Provider: Anthropic. Released 2026-02-17. Access: API.
Aggregate Rating 1739 ± 9, rank #262 of 2168 rated models, from 175 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ClawBench (OpenClaw) | 33.3 | Success rate (%) over the 153 ClawBench tasks on 144 live pl | 100 |
| ClawBench (OpenClaw) - Academic | 50 | Success rate (%) on the 14 academic tasks of the 153-task Cl | 100 |
| ClawBench (OpenClaw) - Daily | 44.2 | Success rate (%) on the 52 daily-life tasks of the 153-task | 100 |
| ClawBench (OpenClaw) - Social | 38.9 | Success rate (%) on the 18 social tasks of the 153-task Claw | 100 |
| HypoArena | 1654 | Average of the six domain ratings: Bradley-Terry-Davidson ra | 100 |
| HypoArena - Financial Analysis | 1739.2 | Rating on the 114 financial analysis cases: Bradley-Terry-Da | 100 |
| HypoArena - IT Operations | 1699.8 | Rating on the 146 IT operations cases: Bradley-Terry-Davidso | 100 |
| HypoArena - Safety Investigation | 1721.6 | Rating on the 103 safety investigation cases: Bradley-Terry- | 100 |
| NVIDIA ComputeEval - cuBLAS | 98.8 | Pass@1 (%, zero-shot, 2026.1 release, 81 problems) | 100 |
| DeepResearchBench | 54.9 | Average Score | 97.5 |
| Horangi 4 - Ko-MT-Bench | 89.5 | Judge rating (1-10, x10) | 95.2 |
| NVIDIA ComputeEval - Math Libs | 91.1 | Pass@1 (%, zero-shot, 2026.1 release, 101 problems) | 94.7 |
Interactive version: theaggregate.ai/model?slug=claude-sonnet-4-6-high · How It Works · Data refreshed daily, snapshot 2026-10-20.