DiscoverPhysics: leaderboard
Metric: Mean Explanation Score (self-reported). Source: benchmarklist.com. 11 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.7 | 61 |
| 2 | GPT-5.5 | 59 |
| 3 | Qwen 3.5 397B A17B | 33 |
| 4 | GPT-5.4 | 29 |
| 5 | Kimi K2.5 | 29 |
| 6 | Claude Sonnet 4.6 | 28 |
| 7 | Qwen 3 235B A22B | 24 |
| 8 | Claude Haiku 4.5 | 23 |
| 9 | GPT-OSS-120B | 22 |
| 10 | GPT-OSS-20B | 14 |
| 11 | Llama 3.3 70B | 10 |
Interactive version: theaggregate.ai/benchmark?slug=discoverphysics · How It Works · Data refreshed daily, snapshot 2026-09-05.