Django Benchmark Suite - Program Repair: leaderboard
Metric: pass@1 (%; Program repair: the model fixes a broken method from the file and the Django test stack trace; the output is inserted into the file and counts only when every Django test mapped to the method passes; 359 held-out methods of a leakage-controlled Django repository snapshot; greedy decoding, no agent scaffold). Source: arxiv.org. Saturation forecast: Around December 2026. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 79.67 |
| 2 | Claude Opus 4.5 | 78.27 |
| 3 | Claude Sonnet 4.5 | 69.36 |
| 4 | Gemini 2.5 Pro | 67.97 |
| 5 | Gemini 3 Flash | 62.95 |
| 6 | GPT-4.1 | 61.84 |
| 7 | Claude Haiku 4.5 | 61.56 |
| 8 | Gemini 3 Pro | 61 |
| 9 | Qwen 2.5 Coder 32B Instruct | 47.35 |
| 10 | Qwen 3 32B | 46.8 |
| 11 | Qwen 2.5 Coder 7B Instruct | 17.83 |
Interactive version: theaggregate.ai/benchmark?slug=django-benchmark-suite-program-repair · How It Works · Data refreshed daily, snapshot 2026-09-29.