SeePhys: leaderboard
SeePhys — physics reasoning from diagrams spanning mechanics to modern physics; multimodal problem solving.
Metric: Total Accuracy (%). Source: seephys.github.io. Saturation forecast: Around January 2027. 22 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 (High) | 63.2 | #98 (GPT-5) |
| 2 | Gemini 2.5 Pro | 54.9 | #152 |
| 3 | O4 Mini | 51.9 | #180 |
| 4 | Grok 4 | 49 | #181 |
| 5 | O1 | 45.6 | #183 |
| 6 | Doubao-1.5-Pro | 43.9 | #287 |
| 7 | O3 Mini | 40.3 | #272 |
| 8 | GPT-4.1 | 35.3 | #248 |
| 9 | Claude 3.7 Sonnet | 34.6 | #249 |
| 10 | QVQ-72B-Preview | 22.5 | #440 |
| 11 | GPT-4o | 21.9 | #343 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=seephys · How It Works · Data refreshed daily, snapshot 2026-10-20.