VMLU - Social Science — leaderboard

Metric: Accuracy (%). Source: vmlu.ai. 25 models tracked.

Top models

#ModelScore
1QwQ-32B78.49
2Llama 3 70B74.91
3GPT-471.78
4GPT-4o Mini70.95
5QwQ 32B-Preview69.84
6Gemma 2 9B (IT)65.79
7Qwen 2.5 7B Instruct63.27
8gemma-7B (IT)44.93
9Phi-3-small-128k-instruct44.82
10Phi-3-small-8k-instruct43.6
11Qwen 2 7B Instruct35.24
12Qwen-7B35.07
13gemma-2B (IT)29.59
14bloom-7b126.26
15bloom-1b725.09

Interactive version: theaggregate.ai/benchmark?slug=vmlu-social-science · How the rankings work · Data refreshed daily, snapshot 2026-07-22.