Real World Tasks on HealthBench Hard
38.6ScoreGPT-5-mini High
Evaluation Results
| Method | Links | |
|---|---|---|
| GPT-5-mini High2026.06 | 38.6 | |
| GPT-5.2 High2026.06 | 36.6 | |
| Seed2.0 Pro2026.06 | 28.3 | |
| Gemini-3-Flash High2026.06 | 21.5 | |
| Seed2.0 Lite2026.06 | 20 | |
| Gemini-3-pro High2026.06 | 15 | |
| Claude-Opus-4.52026.06 | 11 | |
| Claude-Sonnet-4.52026.06 | 10.9 |