HealthBench
Open-ended health conversations graded against physician-written rubrics.
- 1gpt-oss-120b · OpenAI57.6 score
- 2GPT-6 Luna · OpenAI54.5 score
- 3gpt-oss-20b · OpenAI42.5 score
How each maker reported it
- gpt-oss-120b Figures 1, 2, 4; Table 2
OpenAI's own pass@1 results at reasoning level high. AIME and HLE with tools; Codeforces with a terminal tool. o4-mini and o3 values read from OpenAI's charts.
- GPT-6 Luna Appendix A, p.120–154; GPT-6.1 Sol addendum Table 9
The card reports safety and Preparedness evaluations, not general capability benchmarks. Cyber and bio scores are at maximum reasoning effort; HealthBench scores are length-adjusted. MentalHealthBench comes from the GPT-6.1 Sol addendum.
- gpt-oss-20b Figures 1, 2, 4; Table 2
OpenAI's own pass@1 results at reasoning level high. AIME and HLE with tools; Codeforces with a terminal tool. o3-mini was run without tools on AIME, HLE and Codeforces, so those cells are left empty. Baseline values read from OpenAI's charts.
Only each model's own maker's number is shown; scores a maker reports for rival models are left out.