Science Leaderboard — leaderboard
Metric: Average Accuracy (%). Source: huggingface.co. 34 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4 Turbo | 67 |
| 2 | Llama 3 70B Instruct | 61.9 |
| 3 | MAmmoTH2-8x7B-Plus | 55.5 |
| 4 | Llama 3 8B Instruct | 45.4 |
| 5 | internlm2-20B | 43.3 |
| 6 | Mixtral 8x7B Instruct | 42.6 |
| 7 | Yi 34B (Base) | 39.9 |
| 8 | deepseek-llm-67B-base | 39.4 |
| 9 | internlm2-7B | 31.2 |
| 10 | Yi 6B (Base) | 23.6 |
| 11 | Llama 2 13B Base | 22.9 |
| 12 | deepseek-llm-7B-base | 21.7 |
Interactive version: theaggregate.ai/benchmark?slug=science-leaderboard · How the rankings work · Data refreshed daily, snapshot 2026-07-22.