HELM MMLU - College Chemistry: leaderboard
Metric: EM. Source: crfm.stanford.edu. 79 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 1.5 Pro (002) | 67 |
| 2 | DeepSeek V3 | 65 |
| 3 | Qwen 2 72B Instruct | 65 |
| 4 | Claude 3.5 Sonnet (20241022) | 63 |
| 5 | Nova Pro | 63 |
| 6 | Qwen 2.5 72B Instruct | 62 |
| 7 | Gemini 1.5 Pro (001) | 62 |
| 8 | Qwen 1.5 110B | 62 |
| 9 | GPT-4o (2024-05-13) | 61 |
| 10 | GPT-4o (2024-08-06) | 60 |
| 11 | Llama 3.1 405B Instruct | 60 |
| 12 | Claude 3 Opus (20240229) | 60 |
| 13 | Gemini 1.5 Flash (001) | 60 |
| 14 | Llama 3.1 70B Instruct | 59 |
| 15 | Claude 3.5 Sonnet (20240620) | 59 |
Interactive version: theaggregate.ai/benchmark?slug=helm-mmlu-college-chemistry · How It Works · Data refreshed daily, snapshot 2026-09-08.