Med-CoReasoner - Global-MMLU-Medical: leaderboard
Baseline results from the Med-CoReasoner paper (Gao et al., 2026) on the medical-subject subset of Global-MMLU (1,505 multiple-choice items per language), averaged over Chinese, Japanese, Korean, German, French, Spanish, Italian and Swahili. Exact-match accuracy (%) with sampling temperature 0.7 and low reasoning effort for reasoning models; higher is better.
Metric: Accuracy (%). Source: arxiv.org. Saturation forecast: Estimated already saturated. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 (Low) | 84.92 | #172 (GPT-5.2) |
| 2 | GPT-5.1 (Low) | 84.11 | #225 (GPT-5.1) |
| 3 | GPT-4o | 81.79 | #741 |
| 4 | DeepSeek V3.2 | 79.59 | #403 |
| 5 | Qwen 2.5 72B Instruct | 74.52 | #1093 |
| 6 | Llama 3.1 70B Instruct | 73.63 | #1541 |
| 7 | Qwen 2.5 32B Instruct | 71.89 | #1216 |
| 8 | Claude 3.5 Haiku | 68.21 | #1563 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=med-coreasoner-global-mmlu-medical · How It Works · Data refreshed daily, snapshot 2026-11-12.