Claude Opus 5 (Medium): benchmark results
Provider: Anthropic. Released 2026-07-24. Access: API.
Aggregate Rating 1884 ± 11, rank #46 of 2168 rated models, from 118 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| BIABench | 65.4 | Outcome score (0-100): mean over 16 end-to-end bioimage-anal | 100 |
| Chinese Competitive Debating - Winner Prediction | 66.2 | Accuracy (%; the model reads a full match transcript and giv | 100 |
| Korean LLM Eval - CLIcK | 95.84 | Accuracy (%, 0-shot) | 100 |
| Korean LLM Eval - HAE-RAE Bench | 95.84 | Accuracy (%, 0-shot) | 100 |
| Korean LLM Eval - KMMLU-HARD | 88.4 | Accuracy (%, 0-shot) | 100 |
| Korean LLM Eval - KMMLU-Pro | 95.64 | Accuracy (%, 0-shot) | 100 |
| Korean LLM Eval - KoBALT-700 | 88.14 | Accuracy (%, 0-shot) | 100 |
| Korean LLM Eval - MuSR (Ko) | 86.53 | Accuracy (%, 0-shot) | 100 |
| AA-Omniscience Index - Software Engineering (SWE) - R | 64 | Omniscience Index | 98.8 |
| AA-Omniscience Index - Software Engineering (SWE) - Swift | 88 | Omniscience Index | 98 |
| AA-Omniscience Index - Software Engineering (SWE) - Dart | 76 | Omniscience Index | 97.4 |
| AA-Omniscience Index - Software Engineering (SWE) - Kotlin | 82 | Omniscience Index | 97.4 |
Interactive version: theaggregate.ai/model?slug=claude-opus-5-medium · How It Works · Data refreshed daily, snapshot 2026-10-20.