Merge-Bench: leaderboard
This paper applies machine learning to the difficult and important task of version control merging. (1) We constructed a dataset, Merge-Bench, of 7938 real-world merge conflict hunks from 1439 GitHub repositories. The ground truth is the merge resolution that developers committed to the repository. Our dataset construction methodology is scalable to arbitrary amounts of data since no manual labeli
Metric: Equivalent text (self-reported). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 18 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 54.7 | #152 |
| 2 | O3 Pro | 46.1 | #116 |
| 3 | Claude Opus 4 | 44.4 | #163 |
| 4 | Grok 4 | 33.4 | #181 |
| 5 | QwQ-32B | 32.1 | #421 |
| 6 | Qwen 3 235B A22B | 30.9 | #313 |
| 7 | Llama 4 Maverick | 26.2 | #464 |
| 8 | Qwen 3 32B | 13.2 | #443 |
| 9 | Qwen 3 14B | 12.9 | #537 |
| 10 | Qwen 3 8B | 5.5 | #674 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=merge-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.