LLM Stats (NL2Repo): leaderboard
Natural-language-to-repository benchmark evaluating whether coding agents can implement repo-level changes from high-level task descriptions.
Metric: Score (%). Source: llm-stats.com. Saturation forecast: Around January 2027. 25 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4.1 Flash (1017) | 68.2 | #20 |
| 2 | DeepSeek V4.1 Flash | 64 | #36 |
| 3 | DeepSeek V4 Pro (0813) | 61.5 | #53 |
| 4 | MiniMax-M3.1-Flash | 61.2 | #33 |
| 5 | Hy4 preview | 58.9 | #38 |
| 6 | GLM-5.3 | 58 | #39 |
| 7 | DeepSeek-V4-Flash-Vision-Exp | 57.7 | #76 |
| 8 | GLM-5.3 Flash | 56.3 | #65 |
| 9 | Qwen 3.8 Max | 55.9 | #35 |
| 10 | DeepSeek V4 Flash (0731) | 54.2 | #78 |
| 11 | GLM-5.2 | 48.9 | #73 |
| 12 | Qwen3.8-Flash-Next | 48.1 | #72 |
| 13 | Qwen 3.7 Max | 47.2 | #75 |
| 14 | Seed 2.1 Pro | 47 | #67 |
| 15 | Hy3 | 45.6 | #93 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=llm-stats-nl2repo · How It Works · Data refreshed daily, snapshot 2026-10-20.