Claude Sonnet 4.6: healthcare benchmark results
Anthropic · 4 boards · snapshot reviewed September 28, 2026
The index currently holds 4 results for Claude Sonnet 4.6: 0.442 on HealthBench Professional (25 of 29 indexed rows), 23.0 ± 2.6 on PhysicianBench (14 of 21 indexed rows), 0.36 on EHR-Complex (13 of 18 indexed rows), 67.1% on WHBench (2 of 22 indexed rows). Scores below sit on different scales and come from different graders, so read each against its own benchmark, never against the others.
Results by benchmark
| benchmark | score | index position | as of |
|---|---|---|---|
| HealthBench Professional length-adjusted; Anthropic protocol: adaptive thinking at max effort, Claude Opus 4.8 grader, averaged over 5 trials, no tools or custom system prompt; comparison column in the Sonnet 5 card. Other Anthropic prints: 44.4% (Fable card Figure 8.18.2.A), 41.7% (Opus 4.8 card, Sonnet 4.6 grader) | 0.442 | 25 of 29 | 2026-06 |
| PhysicianBench via paper, arxiv.org | 23.0 ± 2.6 | 14 of 21 | 2026-05 |
| EHR-Complex validation configuration via paper, arxiv.org | 0.36 | 13 of 18 | 2026-06 |
| WHBench 95% CI 64.5-69.6 via paper, arxiv.org | 67.1% | 2 of 22 | 2026-03 |
Position follows the rows held in this index and is not a controlled comparison. Sources may use different graders or task subsets. Positions include separate configurations. A source board may contain additional rows; its size is listed separately when it differs. Config caveats, where a source noted any, are on each benchmark's page, and the document behind each score is cited on the sources page.
Which healthcare benchmarks is Claude Sonnet 4.6 scored on?
As of September 28, 2026, Claude Sonnet 4.6 has indexed results on 4 tracked benchmarks: HealthBench Professional, PhysicianBench, EHR-Complex, WHBench.
Which results are indexed for Claude Sonnet 4.6?
Claude Sonnet 4.6 stands at 0.442 on HealthBench Professional (25 of 29 indexed rows), 23.0 ± 2.6 on PhysicianBench (14 of 21 indexed rows), 0.36 on EHR-Complex (13 of 18 indexed rows), 67.1% on WHBench (2 of 22 indexed rows).
The benchmarks themselves are described on their pages, linked in the table above, and the whole field is on the index.