Official benchmarks map / MMLU-Pro
internlm/Intern-S2-Preview-397B is ranked first on the MMLU-Pro leaderboard with 89.75, 1.8% ahead of FINAL-Bench/Darwin-180B-RSI. Read live from the Hugging Face API on November 8, 2026.
| # | Model | Organization | Score |
|---|---|---|---|
| 1 | internlm/Intern-S2-Preview-397B | internlm | 89.75 |
| 2 | FINAL-Bench/Darwin-180B-RSI | FINAL-Bench | 88.12 |
| 3 | MiniMaxAI/MiniMax-M2.1 | MiniMaxAI | 88 |
| 4 | internlm/Intern-S2-Preview | internlm | 88 |
| 5 | Qwen/Qwen3.5-397B-A17B | Qwen | 87.8 |
| 6 | deepseek-ai/DeepSeek-V4-Pro | deepseek-ai | 87.5 |
| 7 | moonshotai/Kimi-K2.5 | moonshotai | 87.1 |
| 8 | nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 | nvidia | 86.8 |
| 9 | nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-NVFP4 | nvidia | 86.8 |
| 10 | Qwen/Qwen3.5-122B-A10B | Qwen | 86.7 |
MMLU-Pro is a dataset that Hugging Face marks as an official benchmark. Its leaderboard is built automatically from the .eval_results files that model repositories publish, so entries are mostly self-reported by model authors and evaluation settings differ between submissions. A verified tag appears only where Hugging Face marks an entry as verified.
Original board: huggingface.co/datasets/TIGER-Lab/MMLU-Pro
Part of the Hugging Face official benchmarks map by QuantID. Unofficial. Not affiliated with or endorsed by Hugging Face.