VMLU - STEM — leaderboard
Metric: Accuracy (%). Source: vmlu.ai. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | QwQ-32B | 81.11 |
| 2 | GPT-4 | 63.84 |
| 3 | QwQ 32B-Preview | 63.42 |
| 4 | Llama 3 70B | 61.7 |
| 5 | GPT-4o Mini | 58 |
| 6 | Gemma 2 9B (IT) | 56.31 |
| 7 | Qwen 2.5 7B Instruct | 55.43 |
| 8 | gemma-7B (IT) | 39.95 |
| 9 | Phi-3-small-128k-instruct | 39.31 |
| 10 | Phi-3-small-8k-instruct | 38.72 |
| 11 | Qwen-7B | 30.64 |
| 12 | bloom-1b7 | 25.13 |
| 13 | bloom-7b1 | 25.08 |
| 14 | gemma-2B (IT) | 24.39 |
| 15 | falcon-7B | 24.19 |
Interactive version: theaggregate.ai/benchmark?slug=vmlu-stem · How the rankings work · Data refreshed daily, snapshot 2026-07-22.