Claude Opus 4.7 (Thinking): benchmark results
Claude Opus 4.7 evaluated with thinking enabled. Provider: Anthropic. Released 2026-04-16. Access: API.
Aggregate Rating 1749 ± 16, rank #242 of 2168 rated models, from 51 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ATRBench - Always Ask | 32.1 | ATRBench test-session accuracy TSAcc (%, 0-100): the share o | 100 |
| Chinese Classical Bench - Char-Gloss Judge | 73.6 | Score (%) | 100 |
| Chinese Classical Bench - Fill-In Exact | 88 | Score (%) | 100 |
| Chinese Classical Bench - Translate Judge | 80.2 | Score (%) | 100 |
| HierSVA-B - Compile Rate | 89.2 | Module compile rate (%, C1): share of the 342 modules whose | 100 |
| KnotBench | 54.6 | Accuracy (%) (self-reported) | 100 |
| KnotBench - Crossing Count | 14 | Accuracy (%) on the 100 C0 items: how many crossings a rende | 100 |
| KnotBench - PD Code Choice | 35.5 | Accuracy (%) on the 200 D1 items: which of four PD codes des | 100 |
| KnotBench - Reidemeister Move (Images) | 32.5 | Accuracy (%) on the 200 B0-I items: which Reidemeister move | 100 |
| KnotBench - Same Chirality (PD Codes) | 92 | Accuracy (%) on the 100 A1-S items: whether two PD codes of | 100 |
| KnotBench - Same Knot (Images) | 55 | Accuracy (%) on the 200 A0-I items: whether two rendered dia | 100 |
| SecRespond - Vulnerability Risk Planning | 72.6 | Planning CAP-score (%) on the Vulnerability Risk (VUL) capab | 100 |
Interactive version: theaggregate.ai/model?slug=claude-opus-4-7-thinking · How It Works · Data refreshed daily, snapshot 2026-10-20.