OpenCompass Knowledge - Engineering — leaderboard

Metric: Score (%). Source: rank.opencompass.org.cn. 23 models tracked.

Top models

#ModelScore
1GPT-5.4 Mini (High)96.2
2GPT-5.4 (High)96.2
3Gemini 3.1 Pro (Preview)95.4
4Claude Opus 4.7 (High)95
5Claude Sonnet 4.6 (High)94.2
6Qwen 3.5 397B A17B93.3
7Qwen 3.6 Max Preview92.5
8Gemini 3.1 Flash Lite (Preview)91.7
9DeepSeek V4 Pro91.2
10DeepSeek V4 Flash91.2
11GLM-5.190.8
12Gemma 4 31B (IT)90
13Hy3-preview (High)90
14Step 3.5 Flash89.6
15Kimi K2.689.2

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