Django Benchmark Suite - Method Completion: leaderboard
Metric: pass@1 (%; Method completion: the model completes a method from its first half and the surrounding file, without a docstring; 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 | Gemini 3 Pro | 78.27 |
| 2 | GPT-5 | 77.99 |
| 3 | Claude Sonnet 4.5 | 77.44 |
| 4 | Claude Opus 4.5 | 76.6 |
| 5 | Gemini 3 Flash | 73.26 |
| 6 | GPT-4.1 | 64.9 |
| 7 | Qwen 2.5 Coder 32B Instruct | 60.17 |
| 8 | Claude Haiku 4.5 | 58.77 |
| 9 | Gemini 2.5 Pro | 51.25 |
| 10 | Qwen 2.5 Coder 7B Instruct | 50.97 |
| 11 | Qwen 3 32B | 43.45 |
Interactive version: theaggregate.ai/benchmark?slug=django-benchmark-suite-method-completion · How It Works · Data refreshed daily, snapshot 2026-09-29.