OpenCompass LLM - Math — leaderboard

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

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)76.5
2Kimi K2.676.4
3DeepSeek V4 Pro74.8
4Hy3-preview (High)74.5
5Qwen 3.6 Max Preview73.4
6Claude Sonnet 4.6 (High)72.4
7Claude Opus 4.7 (High)72.4
8GPT-5.4 (High)72.1
9Qwen 3.5 397B A17B70.9
10DeepSeek V4 Flash69.2
11Step 3.5 Flash68.7
12GLM-5.167.9
13Gemma 4 31B (IT)66.5
14Gemini 3.1 Flash Lite (Preview)66.4
15GPT-5.4 Mini (High)66.1

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