SuperCLUE General (May 2026) - Math Reasoning: leaderboard

Metric: Score. Source: www.superclueai.com. 24 models tracked.

Top models

#ModelScore
1DeepSeek V4 Flash (Max)82.69
2Gemini 3.5 Flash (High)82.46
3Gemini 3.1 Pro (Preview) (High)82.46
4GPT-5.5 (High)82.46
5Claude Opus 4.7 (High)80.7
6Claude Opus 4.8 (High)78.95
7Kimi K2.6 (Thinking)75.93
8Gemma 4 31B75.44
9DeepSeek V4 Pro (Max)71.93
10GLM-5.170.18
11MiMo-V2.5-Pro69.64
12Ernie 5.168.42
13Spark X268.42
14Step 3.7 Flash66.67
15Qwen 3.6 Max (Preview) (Thinking)66.67

Interactive version: theaggregate.ai/benchmark?slug=superclue-general-may-2026-math-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-19.