Claude Opus 4.6 (Non-reasoning): benchmark results
Claude Opus 4.6 evaluated with reasoning disabled. Provider: Anthropic. Released 2026-02-05. Access: API.
Aggregate Rating 1748 ± 14, rank #245 of 2168 rated models, from 52 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ComboShoppingBench - Claim Faithfulness | 94.8 | Pass rate (%; LLM-judged) | 100 |
| HINTBench - Fine Risk-Step Localization (Strict) | 37.93 | Strict F1 (%) for fine risk-step localization: a predicted r | 100 |
| RuleWeaver - Cross-Source - Rubric Score | 50.27 | Judge rubric score (0-100) | 100 |
| RuleWeaver - Cross-Source - Rule Recall | 62.5 | Gold-rule recall (%) | 100 |
| SEAL - MASK | 96.28 | Score | 100 |
| o11y-bench - Pass@3 | 90.48 | Tasks passed on at least one of three attempts, Pass@3 (%) | 94.7 |
| PRBench Finance | 53.28 | Score (%) | 91.2 |
| ComboShoppingBench - Semantic Requirements | 76.6 | Pass rate (%; LLM-judged) | 90.5 |
| Professional Reasoning Bench - Legal | 52.27 | Score (%) | 89.5 |
| SEAL - Professional Reasoning Benchmark - Finance | 53.28 | Score | 89.2 |
| SEAL - Professional Reasoning Benchmark - Legal | 52.27 | Score | 87.8 |
| o11y-bench - Pass^3 | 66.67 | Tasks passed on all three attempts, Pass^3 (%) | 86.8 |
Interactive version: theaggregate.ai/model?slug=claude-opus-4-6-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-10-20.