SuperCLUE General (2025 Annual) - Science Reasoning: leaderboard
Metric: Score. Source: www.superclueai.com. 34 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 78.63 |
| 2 | Gemini 3.1 Pro (Preview) (High) | 77.05 |
| 3 | GPT-5.2 (High) | 75.21 |
| 4 | Doubao-Seed-2.0-Pro-260215 | 75.21 |
| 5 | Qwen 3.5 122B A10B (Thinking) | 74.59 |
| 6 | Gemini 3 Flash (Preview) | 74.17 |
| 7 | Gemini 3 Pro (Preview) | 73.77 |
| 8 | Claude Opus 4.5 (Thinking) | 73.77 |
| 9 | Qwen 3.5 397B A17B (Thinking) | 73.77 |
| 10 | Qwen 3.5 27B (Thinking) | 72.13 |
| 11 | DeepSeek V3.2 (Thinking) | 71.31 |
| 12 | Qwen 3 Max (Thinking) | 68.85 |
| 13 | Grok 4 | 68.03 |
| 14 | GLM-5 | 67.21 |
| 15 | Kimi K2.5 (Thinking) | 67.21 |
Interactive version: theaggregate.ai/benchmark?slug=superclue-general-2025-annual-science-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-19.