CL4SE - Code Review (5-Shot): leaderboard
Metric: Accuracy (%) of the accept-or-reject decision on 1,916 GitHub pull requests from 32 repositories (1,191 and 725 of the two outcomes) given the code change and its multi-turn review conversation; five same-repository review examples with their full multi-turn conversations in context; temperature 0; CL4SE (context learning for software engineering); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 5 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Max | 76.88 | #201 |
| 2 | Claude 3.5 Haiku (20241022) | 67.72 | #574 |
| 3 | GPT-OSS-120B | 66.39 | #330 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=cl4se-code-review-5-shot · How It Works · Data refreshed daily, snapshot 2026-10-11.