Claude Opus 5 (Max): benchmark results
Provider: Anthropic. Released 2026-07-24. Access: API.
Unified ELO 1771 ± 1, rank #21 of 1919 rated models, from 105 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Agent Arena - Bash Recovery | 15.69 | Bash Recovery (%) | 100 |
| Chatbot Arena (Text - Chinese) | 1569 | Arena Score | 100 |
| Conceptual Reasoning Index - Argument Evaluation (LMCA) | 63.3 | Chance-Corrected Score (0-100) | 100 |
| LiveBench AMPS Hard | 99.01 | Score | 100 |
| RSI-Exam | 0.46 | Mean hidden-set score | 100 |
| Toolathlon | 80.6 | Score (self-reported) | 100 |
| Vals AI MedCode | 63.57 | Accuracy (%) | 100 |
| Vals AI Public Benefits Bench | 76.93 | Accuracy (%) | 100 |
| Vals AI Time Horizon Index | 13.67 | Progress (%) | 100 |
| Chatbot Arena (Text - Math) | 1529 | Arena Score | 99.9 |
| Chatbot Arena (Text - Korean) | 1501 | Arena Score | 99.8 |
| Epoch AI - Dtbench | 97.6 | Score | 99.4 |
Interactive version: theaggregate.ai/model?slug=claude-opus-5-max · How It Works · Data refreshed daily, snapshot 2026-09-08.