Vals AI MedQA: leaderboard
USMLE-style medical board exam multiple-choice questions covering clinical reasoning, diagnostics, and broad medical topics from a 2,000-item pool.
Metric: Accuracy (%). Source: www.vals.ai. Saturation forecast: Estimated already saturated. 95 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 (2024-12-17) (High) | 96.52 | #149 (O1 (2024-12-17)) |
| 2 | GPT-5.1 (2025-11-13) (High) | 96.38 | #101 (GPT-5.1 (2025-11-13)) |
| 3 | Gemini 3.1 Pro (Preview) (High) | 96.37 | #58 (Gemini 3.1 Pro (Preview)) |
| 4 | GPT-5 (2025-08-07) (High) | 96.32 | |
| 5 | GPT-5.4 (2026-03-05) (xHigh) | 96.09 | #59 (GPT-5.4 (2026-03-05)) |
| 6 | O3 (2025-04-16) (High) | 96.06 | #121 (O3 (2025-04-16)) |
| 7 | GPT-5 Mini (2025-08-07) (High) | 96.06 | #172 (GPT-5 Mini (2025-08-07)) |
| 8 | Gemini 3 Pro (Preview) (High) | 96.03 | #66 (Gemini 3 Pro (Preview)) |
| 9 | O4 Mini (2025-04-16) (High) | 96.02 | #179 (O4 Mini (2025-04-16)) |
| 10 | Claude Opus 4.5 (20251101) (Thinking) | 95.88 | #74 (Claude Opus 4.5 (20251101)) |
| 11 | Gemini 3 Flash (Preview) (High) | 95.81 | #82 (Gemini 3 Flash (Preview)) |
| 12 | Claude Opus 4.6 (Thinking) | 95.41 | #63 (Claude Opus 4.6) |
| 13 | Qwen 3.5 Plus (Thinking) | 95.21 | #125 (Qwen 3.5 Plus) |
| 14 | O3 Mini (2025-01-31) (High) | 94.83 | #240 (O3 Mini (2025-01-31)) |
| 15 | Claude Sonnet 4.5 (Thinking) | 94.71 | #142 (Claude Sonnet 4.5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=vals-ai-medqa · How It Works · Data refreshed daily, snapshot 2026-10-20.