MMLU-by-task - College Physics — leaderboard
Metric: Accuracy (%). Source: huggingface.co. 1257 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | WizardCoder-Python-34B-V1.0 | 42.16 |
| 2 | internlm-20B Chat | 42.16 |
| 3 | Mistral-7B-v0.1 | 39.22 |
| 4 | StableBeluga2 | 39.22 |
| 5 | LLaMA-65B | 38.24 |
| 6 | CodeLlama-7B-hf | 38.24 |
| 7 | falcon-180B | 38.24 |
| 8 | Llama 2 70B Base | 37.25 |
| 9 | mpt-30B Instruct | 35.29 |
| 10 | CodeLlama-7B-Python-hf | 35.29 |
| 11 | Llama 2 70B Chat (HF) | 33.33 |
| 12 | Llama 2 70B Chat GPTQ | 33.33 |
| 13 | CodeLlama-13B-Instruct-hf | 32.35 |
| 14 | Llama 2 13B Chat Base | 31.37 |
| 15 | opt-1.3B | 31.37 |
Interactive version: theaggregate.ai/benchmark?slug=mmlu-by-task-college-physics · How the rankings work · Data refreshed daily, snapshot 2026-07-22.