RealBench (RAG) - Level 4: leaderboard
RealBench (2026) asks LLMs to build a whole Python repository from natural-language requirements and UML system designs, here with retrieval-augmented generation (file by file, each conditioned on the earlier generated files it imports) on the level-4 repositories (about 2,360 lines of code). The score is the percentage of the original unit tests that the generated repository passes, higher is better.
Metric: Test pass rate (%), passed tests over all tests of each task, averaged over the level-4 tasks (10 repositories of about 2,360 lines), with retrieval-augmented generation (file by file, each conditioned on the earlier generated files it imports); RealBench repository-level Python code generation from natural-language requirements plus UML package and class diagrams (61 real repositories in four size levels), one greedy solution per task through the official APIs; higher is better. Source: arxiv.org. Saturation forecast: Around 2031. 6 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4 (20250514) | 21.33 | #405 |
| 2 | Qwen 2.5 Coder 7B Instruct | 15.11 | #3914 |
| 3 | DeepSeek V3 | 13.27 | #729 |
| 4 | GPT-4o (2024-05-13) | 9.95 | #732 |
| 5 | Gemini 2.5 Flash | 8.01 | #463 |
| 6 | Qwen 3 235B A22B | 0.73 | #628 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=realbench-rag-level-4 · How It Works · Data refreshed daily, snapshot 2026-11-21.