Claude Opus 4.5 (20251101): benchmark results
November 1, 2025 Claude Opus 4.5 snapshot, tracked when sources report the dated API model. Provider: Anthropic. Released 2025-11-24. Access: API.
Aggregate Rating 1787 ± 9, rank #74 of 1656 rated models, from 239 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AraGen - Safety | 91.09 | 3C3H Score (%) | 100 |
| DISBench | 55 | Answer F1 (%) | 100 |
| DISBench (2B Retriever) | 55.5 | Set-level answer F1 (%) on all of DISBench's 122 deep image | 100 |
| DISBench (2B Retriever) - Exact Match | 32 | Exact match (%, the returned set equals the target set) on a | 100 |
| DISBench (2B Retriever) - Inter-Event | 53.4 | Set-level answer F1 (%) on the 65 inter-event queries of DIS | 100 |
| DISBench (2B Retriever) - Intra-Event | 57.9 | Set-level answer F1 (%) on the 57 intra-event queries of DIS | 100 |
| DISBench - Exact Match | 28.7 | Exact Match (%) | 100 |
| DISBench - Inter-Image | 50.7 | Answer F1 (%) | 100 |
| DISBench - Intra-Image | 60 | Answer F1 (%) | 100 |
| HumanAgencyBench - Ask Clarifying Questions | 90.88 | HAB dimension score (0-100, o3-judged rubric) | 100 |
| HumanAgencyBench - Defer Important Decisions | 88.2 | HAB dimension score (0-100, o3-judged rubric) | 100 |
| HumanAgencyBench - Encourage Learning | 78.84 | HAB dimension score (0-100, o3-judged rubric) | 100 |
Interactive version: theaggregate.ai/model?slug=claude-opus-4-5-20251101 · How It Works · Data refreshed daily, snapshot 2026-10-20.