Calibration
SWE-bench Pro → DeepSWE
DeepSWE is estimated from SWE-bench Pro with a Michaelis–Menten + offset curve fitted on 25 models measured on both: y = 0.0000 + 2.0000·x / (1.68402 + x), R² = 0.30, cross-validated error 11.2 pp. It is used for 10 estimates.
| Estimated model | SWE-bench Pro | DeepSWE | Source |
|---|---|---|---|
| Claude Mythos 5 | 80.3% | 64.6% | estimated ± 11.2 pp, low confidence |
| Atria Dawn Preview | 59.6% | 52.3% | estimated ± 11.2 pp, low confidence |
| Claude Opus 4.6 | 53.4% | 48.2% | estimated ± 11.2 pp, low confidence |
| MAI-Thinking-1 | 52.8% | 47.7% | estimated ± 11.2 pp, low confidence |
| Laguna M.1 | 49.2% | 45.2% | estimated ± 11.2 pp, low confidence |
| Laguna XS 2.1 | 47.6% | 44.1% | estimated ± 11.2 pp, low confidence |
| Laguna XS.2 | 46.3% | 43.1% | estimated ± 11.2 pp, low confidence |
| LongCat-Flash-Lite-Sparse | 40.6% | 38.9% | estimated ± 11.2 pp, low confidence |
| LLaDA2.2-flash | 30.1% | 30.3% | estimated ± 11.2 pp, low confidence |
| Mellum2.1-12B-A2.5B-Thinking | 28.0% | 28.5% | estimated ± 11.2 pp, low confidence |
Fitted on
| Model | SWE-bench Pro | DeepSWE |
|---|---|---|
| Claude Fable 5.1 | 81.2% | 67.4% |
| Claude Fable 5.2 | 93.0% | 74.3% |
| Claude Opus 5 | 79.2% | 68.8% |
| Claude Opus 5.5 | 89.9% | 74.2% |
| Claude Sonnet 5.5 | 81.3% | 71.0% |
| DeepSeek V4 Flash 0731 | 52.6% | 54.4% |
| DeepSeek V4 Pro | 55.4% | 62.7% |
| GPT-5.6 Luna | 62.7% | 67.2% |
| GPT-5.6 Sol | 64.6% | 72.7% |
| GPT-5.6 Terra | 63.4% | 69.6% |
| Grok 4.5 | 64.7% | 53.0% |
| Hy4 preview | 65.7% | 64.3% |
| Laguna S 2.1 | 59.4% | 40.4% |
| Ling 3.1 Flash | 68.7% | 59.7% |
| Muse Spark 1.1 | 61.5% | 53.3% |
| Ornith-1.5-35B-A3B | 59.6% | 22.0% |
| Ornith-1.5-397B | 65.1% | 56.0% |
| Qwen3.8-27B | 61.7% | 42.2% |
| Qwen3.8-Flash-Next | 62.5% | 58.7% |
| Qwen3.8 Max | 67.7% | 56.6% |
| Qwen3.8-Omni-Flash | 63.3% | 57.8% |
| Beam 1.0 501B-A23B | 66.1% | 43.3% |
| Beam 1.0 501B-A23B FP8 | 65.9% | 43.0% |
| Beam 1.0 501B-A23B NVFP4 | 65.6% | 42.7% |
| Ripple-65B Preview | 61.4% | 44.3% |