HealthBench: current results
OpenAI · 5,000 conversations · index updated August 16, 2026
Claude Opus 5 holds the top current result on HealthBench, 67.1 as of 2026-08, per OpenAI Deployment Safety Hub (GPT-5.6 system card + August 2026 updates); benchlm.ai and llm-stats.com mirror. 5,000 realistic multi-turn health conversations graded against physician-written rubrics (48,562 criteria) covering accuracy, completeness, context awareness, communication, and instruction following. OpenAI now also reports a length-adjusted variant that penalizes verbosity.
Current results
- 1
Claude Opus 567.1 - 2BBaichuan-M365.1
- 3
GPT-5.2-High63.3 - 4GPT-5.6 Sol57.0
- 5GPT-5.6 Terra57.0
- 6GPT-5.556.5
- 7GPT-5.6 Luna55.8
- 8GPT-5.6 Sol (August)55.0
Result detail
| # | model | score | as of | |
|---|---|---|---|---|
| 1 | | Claude Opus 5 Anthropic raw/unadjusted, Anthropic system-card protocol (max effort, no tools, five trials); length-adjusted 57.8; mirrored on benchlm.ai | 67.1 | 2026-08 |
| 2 | B | Baichuan-M3 Baichuan self-run in Baichuan-M3 paper (arXiv 2602.06570) | 65.1 | 2026-02 |
| 3 | | GPT-5.2-High OpenAI as run by Baichuan in the M3 paper, not OpenAI-reported | 63.3 | 2026-02 |
| 4 | | GPT-5.6 Sol OpenAI length-adjusted, max reasoning effort (55.6 unadjusted), GPT-5.6 system card 2026-07-09 | 57.0 | 2026-06 |
| 5 | | GPT-5.6 Terra OpenAI length-adjusted (58.7 unadjusted), max reasoning effort | 57.0 | 2026-06 |
| 6 | | GPT-5.5 OpenAI length-adjusted (58.4 unadjusted), comparison row in GPT-5.6 system card | 56.5 | 2026-06 |
| 7 | | GPT-5.6 Luna OpenAI length-adjusted (55.4 unadjusted), max reasoning effort | 55.8 | 2026-06 |
| 8 | | GPT-5.6 Sol (August) OpenAI ChatGPT production/Instant deployment setting, length-adjusted (52.1 unadjusted), GPT-5.6 August Updates PDF 2026-08-06 | 55.0 | 2026-08 |
Scores appear exactly as OpenAI Deployment Safety Hub (GPT-5.6 system card + August 2026 updates); benchlm.ai and llm-stats.com mirror publishes them (mixed sources). Cross-source rows are not directly comparable: Anthropic reports a raw score under its own protocol, OpenAI leads with length-adjusted numbers at maximum reasoning effort and reports production Instant settings separately, and Baichuan ran its competitors itself. The benchmark's official grader is GPT-4.1. OpenAI has said the parent set is approaching a noise ceiling for frontier models and points to HealthBench Professional for continued measurement.
About the benchmark
| publisher | OpenAI |
|---|---|
| category | rubric-graded benchmarks |
| released | 2025-05 |
| size | 5,000 conversations |
| scale | 0-100 rubric-point percentage (some sites display 0-1), higher better; length-adjusted and unadjusted variants |
| result basis | mixed sources |
| source | OpenAI Deployment Safety Hub (GPT-5.6 system card + August 2026 updates); benchlm.ai and llm-stats.com mirror |
| last frontier result | 2026-08 |
What is HealthBench?
HealthBench is a rubric-graded benchmark from OpenAI, released 2025-05: 5,000 conversations, scored on a 0-100 rubric-point percentage (some sites display 0-1) scale. 5,000 realistic multi-turn health conversations graded against physician-written rubrics (48,562 criteria) covering accuracy, completeness, context awareness, communication, and instruction following. OpenAI now also reports a length-adjusted variant that penalizes verbosity.
Which model leads HealthBench?
Claude Opus 5 (Anthropic) holds the top current result on HealthBench at 67.1, per OpenAI Deployment Safety Hub (GPT-5.6 system card + August 2026 updates); benchlm.ai and llm-stats.com mirror, as of 2026-08.
Where do the HealthBench numbers come from?
From OpenAI Deployment Safety Hub (GPT-5.6 system card + August 2026 updates); benchlm.ai and llm-stats.com mirror (mixed sources). Cross-source rows are not directly comparable: Anthropic reports a raw score under its own protocol, OpenAI leads with length-adjusted numbers at maximum reasoning effort and reports production Instant settings separately, and Baichuan ran its competitors itself. The benchmark's official grader is GPT-4.1. OpenAI has said the parent set is approaching a noise ceiling for frontier models and points to HealthBench Professional for continued measurement.
The rest of the field is on the index, and how sources qualify is on the methodology page. Model names in the table link to cross-benchmark pages.