Software Engineering Tool Use on SWE-bench Verified
76.2Success RateGemini-3-Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-ProCategory=Frontier (≫8B), Temperature=1, Top-p=1, Number of runs=42026.03 | 76.2 | |
| Claude-4-SonnetCategory=Frontier (≫8B), Temperature=1, Top-p=1, Number of runs=42026.03 | 72.7 | |
| Kimi-K2-0905Category=Frontier (≫8B), Temperature=1, Top-p=1, Number of runs=42026.03 | 69.2 | |
| DeepSeek-V3.2-ExpCategory=Frontier (≫8B), Temperature=1, Top-p=1, Number of runs=42026.03 | 67.8 | |
| GPT-OSS-120BCategory=Frontier (≫8B), Temperature=1, Top-p=1, Number of runs=42026.03 | 62 | |
| Gemini-2.5-ProCategory=Frontier (≫8B), Temperature=1, Top-p=1, Number of runs=42026.03 | 59.6 | |
| SWE-Dev-8BCategory=8B Baselines, Temperature=1, Top-p=1, Number of runs=42026.03 | 19.5 | |
| DIVE-8B (RL)Category=Ours, Temperature=1, Top-p=1, Number of runs=42026.03 | 18.3 | |
| DIVE-8B (SFT)Category=Ours, Temperature=1, Top-p=1, Number of runs=42026.03 | 13.2 | |
| EnvScaler-8BCategory=8B Baselines, Temperature=1, Top-p=1, Number of runs=42026.03 | 11.5 | |
| Qwen3-8B (base)Category=Ours, Temperature=1, Top-p=1, Number of runs=42026.03 | 10.8 | |
| WebExplorer-8BCategory=8B Baselines, Temperature=1, Top-p=1, Number of runs=42026.03 | 7 |