Software Engineering Problem Solving on SWE-Bench hard problem subset
10.67Add. SolvedLeanEvolve
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LeanEvolveModel=Qwen-3.5-27B2026.06 | 10.67 | 60 | |
| LeanEvolveModel=Kimi-K2.52026.06 | 8.67 | 70 | |
| LeanEvolveModel=GPT-5.22026.06 | 8 | 70.67 | |
| LeanEvolveModel=Gemma-4-31B2026.06 | 5.33 | 65.33 | |
| LeanEvolveModel=GLM-52026.06 | 4.67 | 62.67 |