DeepSWE v1.1
Code · 2026-06-14
DeepSWE v1.1 evaluates frontier coding agents on the same 113 original, long-horizon software-engineering tasks as v1, with committed patches graded in clean verifier containers and structured per-test reports. Pass@1 is the pooled pass rate over scored rollout attempts; context-window failures and agent timeouts count as failures, while provider, verifier, and network errors are excluded.
Top models (higher is better)
| Model | Score |
|---|---|
| Muse Wonder Full | 79.0 |
| Muse Wonder | 78.8 |
| Gemini 4 Argon | 77.9 |
| DeepSeek V4.1 Flash 1017 | 77.0 |
| Kimi K3.1 Pro | 75.5 |
| Ember-1 | 75.2 |
| Claude Opus 5.5 | 74.2 |
| DeepSeek V4.1 Flash | 74.2 |
| GPT-6 Astra | 74.1 |
| Gemini 3.8 Flash | 73.8 |
| MiniMax M3.1 Flash | 73.7 |
| Claude Opus 5 | 73.6 |
| GPT-5.6 Sol | 72.7 |
| GPT-6.1 Sol | 72.3 |
| MiMo-V2.6-Pro | 71.9 |
| Grok 4.7 | 71.0 |
| GLM-5.3 | 70.5 |
| Claude Sonnet 5.5 | 70.5 |
| GPT-6 Sol | 70.5 |
| Claude Fable 5 | 70.0 |
| GPT-5.6 Terra | 69.6 |
| Kimi K3 | 69.0 |
| MiMo-V2.6-Flash | 67.9 |
| Muse Spark 1.3 | 67.8 |
| Step 5 Preview | 67.7 |
| Grok 4.6 | 67.5 |
| Claude Fable 5.1 | 67.4 |
| GPT-5.6 Luna | 67.2 |
| GPT-5.5 | 67.0 |
| GPT-6 Luna | 67.0 |
| Gemini 3.7 Flash | 65.5 |
| Nanbeige4.3-33B | 65.5 |
| GLM-5.3 Flash | 63.4 |
| DeepSeek V4 Pro 0813 | 62.7 |
| Muse Spark Open | 60.0 |
| Claude Haiku 5.5 | 59.9 |
| Opus 4.8 | 59.0 |
| Qwen3.8-Flash-Next | 58.7 |
| Muse Spark 1.2 | 54.9 |
| Sonnet 5 | 53.8 |