PRBench (Physics Reproduction) - Code Correctness: leaderboard
Metric: Code correctness score (%) from the expert rubric, averaged over three runs and over the 30 tasks with available results; higher is better. Source: arxiv.org. Saturation forecast: Around July 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.3 Codex | 36 | #68 |
| 2 | Kimi K2.5 | 22 | #139 |
| 3 | DeepSeek V3.2 | 19.3 | #198 |
| 4 | GLM-5 | 18.8 | #137 |
| 5 | MiniMax-M2.7 | 16.3 | #248 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=prbench-physics-reproduction-code-correctness · How It Works · Data refreshed daily, snapshot 2026-10-11.