Calibration
SWE-bench Verified → SWE-bench Pro
SWE-bench Pro is estimated from SWE-bench Verified with a inverse Michaelis–Menten curve fitted on 44 models measured on both: y = 1.22685·(x − 0.1653) / (0.1653 + 2.0000 − x), R² = 0.95, cross-validated error 3.2 pp. It is used for 32 estimates.
| Estimated model | SWE-bench Verified | SWE-bench Pro | Source |
|---|---|---|---|
| Apodex 1.1 | 77.7% | 54.1% | estimated ± 3.2 pp, high confidence |
| BTL-4 | 78.4% | 55.0% | estimated ± 3.2 pp, high confidence |
| Claude 3.5 Sonnet | 49.0% | 23.8% | estimated ± 3.2 pp, high confidence |
| Claude 4.1 Opus | 74.5% | 50.1% | estimated ± 3.2 pp, high confidence |
| Claude 4 Sonnet | 72.7% | 47.9% | estimated ± 3.2 pp, high confidence |
| Claude Haiku 4.5 | 73.3% | 48.6% | estimated ± 3.2 pp, high confidence |
| Claude Sonnet 4.5 | 77.2% | 53.4% | estimated ± 3.2 pp, high confidence |
| DeepSeek V3 | 42.0% | 17.9% | estimated ± 3.2 pp, medium confidence |
| Ember-1 | 92.2% | 74.7% | estimated ± 3.2 pp, high confidence |
| Gemini 2.5 Pro | 63.8% | 38.0% | estimated ± 3.2 pp, high confidence |
| GLM-4.7 | 73.8% | 49.2% | estimated ± 3.2 pp, high confidence |
| GPT-4.1 | 54.6% | 28.8% | estimated ± 3.2 pp, high confidence |
| GPT-4.1 mini | 23.6% | 4.5% | estimated ± 3.2 pp, medium confidence |
| Grok Code Fast 1 | 70.8% | 45.7% | estimated ± 3.2 pp, high confidence |
| Hy3 Preview | 74.4% | 50.0% | estimated ± 3.2 pp, high confidence |
| K-EXAONE 2.0 | 68.2% | 42.7% | estimated ± 3.2 pp, high confidence |
| Kimi K2.5 (Reasoning) | 76.8% | 52.9% | estimated ± 3.2 pp, high confidence |
| MAI-Code-1.1-Flash | 72.6% | 47.8% | estimated ± 3.2 pp, high confidence |
| MiMo-V2-Flash | 73.4% | 48.7% | estimated ± 3.2 pp, high confidence |
| MiMo-V2-Omni | 74.8% | 50.4% | estimated ± 3.2 pp, high confidence |
| MiMo-V2-Pro | 78.0% | 54.4% | estimated ± 3.2 pp, high confidence |
| Mistral Medium 3.5 128B | 77.6% | 53.9% | estimated ± 3.2 pp, high confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 52.8% | 27.2% | estimated ± 3.2 pp, high confidence |
| Nemotron 3 Ultra | 71.9% | 47.0% | estimated ± 3.2 pp, high confidence |
| o3-mini | 49.3% | 24.0% | estimated ± 3.2 pp, high confidence |
| Qwen3.5-122B-A10B | 72.0% | 47.1% | estimated ± 3.2 pp, high confidence |
| Qwen3.5-27B | 72.4% | 47.6% | estimated ± 3.2 pp, high confidence |
| Qwen3.5-35B-A3B | 69.2% | 43.9% | estimated ± 3.2 pp, high confidence |
| Solar Open 2 | 70.4% | 45.2% | estimated ± 3.2 pp, high confidence |
| Solar Pro 4 | 70.6% | 45.5% | estimated ± 3.2 pp, high confidence |
| Ternary Bonsai 2 27B | 60.8% | 34.9% | estimated ± 3.2 pp, high confidence |
| ZAYA1-74B-Preview | 53.2% | 27.5% | estimated ± 3.2 pp, high confidence |