OpenCompass Knowledge - Social Science - Chinese (CompassBench 2407): leaderboard

Metric: Score (%). Source: rank.opencompass.org.cn. 36 models tracked.

Top models

#ModelScore
1GPT-4o (2024-05-13)85.6
2Claude 3.5 Sonnet (20240620)84
3Qwen 2 72B Instruct82.4
4DeepSeek V2 Chat (0628)80.8
5GLM-4 (0520)80
6GPT-4o Mini (2024-07-18)79.2
7Mistral Large 2 (Jul)79.2
8Mistral Large79.2
9Llama 3.1 70B Instruct78.4
10Qwen 1.5 110B Chat77.6
11Moonshot v1 8K76
12Yi Large76
13GLM-4 9B Chat69.6
14Qwen 2 7B Instruct67.2
15Mixtral 8x22B Instruct (v0.1)66.4

Interactive version: theaggregate.ai/benchmark?slug=opencompass-knowledge-social-science-chinese-compassbench-2407 · How It Works · Data refreshed daily, snapshot 2026-09-19.