VMLU - Humanities — leaderboard
Metric: Accuracy (%). Source: vmlu.ai. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | QwQ-32B | 71.78 |
| 2 | Llama 3 70B | 68.74 |
| 3 | GPT-4 | 66.14 |
| 4 | GPT-4o Mini | 65.03 |
| 5 | QwQ 32B-Preview | 63.32 |
| 6 | Gemma 2 9B (IT) | 60.8 |
| 7 | Qwen 2.5 7B Instruct | 58.3 |
| 8 | gemma-7B (IT) | 43.39 |
| 9 | Phi-3-small-8k-instruct | 42.32 |
| 10 | Phi-3-small-128k-instruct | 41.78 |
| 11 | Qwen-7B | 34.15 |
| 12 | Qwen 2 7B Instruct | 33.13 |
| 13 | gemma-2B (IT) | 31.01 |
| 14 | falcon-7B | 26.72 |
| 15 | bloom-1b7 | 26.34 |
Interactive version: theaggregate.ai/benchmark?slug=vmlu-humanities · How the rankings work · Data refreshed daily, snapshot 2026-07-22.