Claude Sonnet 4.6 (Medium): benchmark results
Provider: Anthropic. Released 2026-02-17. Access: API.
Aggregate Rating 1695 ± 14, rank #374 of 2170 rated models, from 62 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| FINESSE-Bench - CFA-like Level 1 | 89.79 | Accuracy (%) on the 1069 CFA-like Level 1 multiple-choice qu | 100 |
| FINESSE-Bench - CFA-like Level 2 | 91.81 | Accuracy (%) on the 293 CFA-like Level 2 multiple-choice ite | 98.3 |
| FINESSE-Bench - Trading TA | 83.54 | Accuracy (%) on the 413 Trading_TA numerical and short-answe | 98.3 |
| FINESSE-Bench - CFA-like Level 3 | 82.39 | Accuracy (%) on the 318 CFA-like Level 3 multiple-choice que | 96.7 |
| FINESSE-Bench - CMT-like Level 2 | 89.24 | Accuracy (%) on the 251 CMT-like Level 2 multiple-choice que | 95 |
| ALE-Bench | 1327.3 | Performance (Self-Refine x1) (imported) | 93.3 |
| PointerBench - Text | 49 | Success rate (%) | 88.6 |
| FINESSE-Bench - CFTe-like Level 1 | 84.64 | Accuracy (%) on the 781 CFTe-like Level 1 multiple-choice qu | 86.7 |
| MMGist - Expert Knowledge | 54.9 | Accuracy (%) on the Expert Knowledge dimension (items kept f | 84.6 |
| BeQu - Experiment 1 - Entailment F1 | 43.2 | Entailment F1 (%) | 84.2 |
| BeQu - Experiment 1 - Entailment Recall | 32.4 | Entailment Recall (%) | 84.2 |
| NVIDIA ComputeEval - Math Libs | 89.1 | Pass@1 (%, zero-shot, 2026.1 release, 101 problems) | 84.2 |
Interactive version: theaggregate.ai/model?slug=claude-sonnet-4-6-medium · How It Works · Data refreshed daily, snapshot 2026-11-05.