SuperCLUE General (November 2025) - Science Reasoning: leaderboard
Metric: Score. Source: www.superclueai.com. 32 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 (High) | 56.76 |
| 2 | GPT-5.1 (High) | 53.98 |
| 3 | Qwen 3 Max | 46.02 |
| 4 | Grok 4 Fast (Reasoning) | 44.25 |
| 5 | Claude Opus 4.5 (Thinking) | 44.25 |
| 6 | Gemini 3 Pro (Preview) | 43.36 |
| 7 | DeepSeek V3.2 Speciale | 43.36 |
| 8 | DeepSeek V3.2 (Thinking) | 42.48 |
| 9 | Claude Sonnet 4.5 (Thinking) | 38.94 |
| 10 | Qwen 3 Max (Preview) (Thinking) | 38.94 |
| 11 | GLM-4.7 | 37.5 |
| 12 | Kimi K2 (Thinking) | 36.28 |
| 13 | GLM-4.6 | 36.28 |
| 14 | DeepSeek V3.2 Exp (Thinking) | 36.28 |
| 15 | O3 (High) | 34.51 |
Interactive version: theaggregate.ai/benchmark?slug=superclue-general-november-2025-science-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-19.