Claude Opus 5 (High): benchmark results
Provider: Anthropic. Released 2026-07-24. Access: API.
Aggregate Rating 1914 ± 9, rank #34 of 2170 rated models, from 227 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Chatbot Arena (Document) | 1516 | Elo | 100 |
| DiG-bench - Games Beaten | 50 | Games beaten (of 70) | 100 |
| Finding the Right Fit - TUA-Bench (OpenHands) | 65.23 | Task score (%), OpenHands harness, high reasoning effort, on | 100 |
| Finding the Right Fit - Terminal-Bench 4 (63 Tasks, OpenHands) | 57.14 | Task score (%), OpenHands harness, high reasoning effort, on | 100 |
| FutureEval | 15.92 | Unified Forecasting Score | 100 |
| GameLogicBench | 52.78 | Solve rate (%; all 72 tasks; one sample per model-scaffold c | 100 |
| GameLogicBench - Combo | 53.57 | Solve rate (%; the 28 Combo tasks (several interacting mecha | 100 |
| GameLogicBench - Repo | 43.48 | Solve rate (%; the 23 Repo tasks (a mechanic inside a real o | 100 |
| OpenCompass Knowledge - Science | 97.5 | Score (%) | 100 |
| OpenCompass Knowledge - Social Science | 94.6 | Score (%) | 100 |
| OpenCompass LLM - Knowledge | 94.1 | Score (%) | 100 |
| OpenCompass LLM - Math | 77.3 | Score (%) | 100 |
Interactive version: theaggregate.ai/model?slug=claude-opus-5-high · How It Works · Data refreshed daily, snapshot 2026-11-05.