SuperCLUE General (May 2026) - Code Generation: leaderboard

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

Top models

#ModelScore
1Claude Opus 4.8 (High)83.58
2Gemini 3.1 Pro (Preview) (High)81.47
3Claude Opus 4.7 (High)79.01
4Kimi K2.6 (Thinking)75.79
5DeepSeek V4 Pro (Max)74.95
6GPT-5.5 (High)72.88
7GLM-5.170.8
8Gemini 3.5 Flash (High)70.66
9MiMo-V2.5-Pro67.52
10DeepSeek V4 Flash (Max)66.75
11Grok 4.366.73
12Qwen 3.6 Max (Preview) (Thinking)66.21
13MiniMax-M365.84
14Gemma 4 31B65.76
15Step 3.7 Flash65.33

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