Software Engineering on SWE-bench Verified (generic score)
72.4Generic ScoreQwen3.5 27b
Evaluation Results
| Method | Links | |
|---|---|---|
| Qwen3.5 27bModel size=27B, Model architecture=Dense2026.04 | 72.4 | |
| Devstral Small 2Model size=24B, Model architecture=Dense2026.04 | 68 | |
| gpt-oss-20bModel size=20B, Model architecture=MoE2026.04 | 60.7 | |
| GLM-4.7-FlashModel size=31B, Model architecture=MoE2026.04 | 59.2 |