MMLU-by-task - Elementary Mathematics: leaderboard
Metric: Accuracy (%). Source: huggingface.co. 1257 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Llama 2 70B Chat | 45.5 |
| 2 | Llama 2 70B Base | 43.39 |
| 3 | Llama 2 70B Chat (HF) | 41.01 |
| 4 | LLaMA-65B | 40.48 |
| 5 | vicuna-33B-v1.3 | 37.57 |
| 6 | Mistral-7B-v0.1 | 37.3 |
| 7 | LLaMA-30B | 37.3 |
| 8 | Llama 2 13B Chat Base | 34.13 |
| 9 | mpt-30B | 32.8 |
| 10 | vicuna-7B-v1.3 | 32.54 |
| 11 | vicuna-13B-v1.5 | 32.54 |
| 12 | mpt-30B Instruct | 32.01 |
| 13 | ChatGLM2 6B | 31.75 |
| 14 | CodeLlama-34B-Python-hf | 31.48 |
| 15 | vicuna-13B | 30.95 |
Interactive version: theaggregate.ai/benchmark?slug=mmlu-by-task-elementary-mathematics · How It Works · Data refreshed daily, snapshot 2026-09-05.