Calibration
SWE-bench Verified → AA-SciCode
AA-SciCode is estimated from SWE-bench Verified with a Michaelis–Menten curve fitted on 34 models measured on both: y = 2.0000·x / (2.51037 + x), R² = 0.75, cross-validated error 3.9 pp. It is used for 41 estimates.
| Estimated model | SWE-bench Verified | AA-SciCode | Source |
|---|---|---|---|
| BTL-4 | 78.4% | 47.6% | estimated ± 3.9 pp, high confidence |
| MiMo-V2-Pro | 78.0% | 47.4% | estimated ± 3.9 pp, high confidence |
| Claude Sonnet 4.5 | 77.2% | 47.0% | estimated ± 3.9 pp, high confidence |
| MiMo-V2-Omni | 74.8% | 45.9% | estimated ± 3.9 pp, high confidence |
| Claude 4.1 Opus | 74.5% | 45.8% | estimated ± 3.9 pp, high confidence |
| Step 3.5 Flash | 74.4% | 45.7% | estimated ± 3.9 pp, high confidence |
| Flint | 74.2% | 45.6% | estimated ± 3.9 pp, high confidence |
| Claude Haiku 4.5 | 73.3% | 45.2% | estimated ± 3.9 pp, high confidence |
| Jais-3-MoVA-36B-A4B | 72.8% | 45.0% | estimated ± 3.9 pp, high confidence |
| Claude 4 Sonnet | 72.7% | 44.9% | estimated ± 3.9 pp, high confidence |
| MAI-Code-1.1-Flash | 72.6% | 44.9% | estimated ± 3.9 pp, high confidence |
| Claude 4 Opus | 72.5% | 44.8% | estimated ± 3.9 pp, high confidence |
| Jais-3-32B | 72.4% | 44.8% | estimated ± 3.9 pp, high confidence |
| Qwen3.5-27B | 72.4% | 44.8% | estimated ± 3.9 pp, high confidence |
| Grok Code Fast 1 | 70.8% | 44.0% | estimated ± 3.9 pp, high confidence |
| Solar Open 2 | 70.4% | 43.8% | estimated ± 3.9 pp, high confidence |
| Jais-3-375B-A23B | 70.2% | 43.7% | estimated ± 3.9 pp, high confidence |
| Jais-3-7B | 70.2% | 43.7% | estimated ± 3.9 pp, high confidence |
| Qwen3-Coder 480B-A35B | 69.6% | 43.4% | estimated ± 3.9 pp, high confidence |
| Qwen3.5-35B-A3B | 69.2% | 43.2% | estimated ± 3.9 pp, high confidence |
| DeepSeek V3.1 | 66.0% | 41.6% | estimated ± 3.9 pp, high confidence |
| Kimi K2 | 65.8% | 41.5% | estimated ± 3.9 pp, high confidence |
| Claude 3.7 Sonnet | 62.3% | 39.8% | estimated ± 3.9 pp, high confidence |
| Ternary Bonsai 2 27B | 60.8% | 39.0% | estimated ± 3.9 pp, high confidence |
| Nemotron 3 Super 100B | 60.5% | 38.8% | estimated ± 3.9 pp, high confidence |
| GLM-4.7-Flash | 59.2% | 38.2% | estimated ± 3.9 pp, high confidence |
| DeepSeek-R1-0528 | 57.6% | 37.3% | estimated ± 3.9 pp, high confidence |
| MiniMax M1 80k | 56.0% | 36.5% | estimated ± 3.9 pp, high confidence |
| GPT-4.1 | 54.6% | 35.7% | estimated ± 3.9 pp, high confidence |
| ZAYA1-74B-Preview | 53.2% | 35.0% | estimated ± 3.9 pp, high confidence |
| o3-mini | 49.3% | 32.8% | estimated ± 3.9 pp, high confidence |
| DeepSeek-R1 | 49.2% | 32.8% | estimated ± 3.9 pp, high confidence |
| Claude 3.5 Sonnet | 49.0% | 32.7% | estimated ± 3.9 pp, high confidence |
| Devstral Small | 46.8% | 31.4% | estimated ± 3.9 pp, high confidence |
| Sarvam 105B | 45.0% | 30.4% | estimated ± 3.9 pp, high confidence |
| Claude 3.5 Haiku | 40.6% | 27.8% | estimated ± 3.9 pp, medium confidence |
| Sarvam 30B | 34.0% | 23.9% | estimated ± 3.9 pp, medium confidence |
| Gemini 2.5 Flash-Lite | 31.6% | 22.4% | estimated ± 3.9 pp, medium confidence |
| Gemini 2.0 Flash | 21.4% | 15.7% | estimated ± 3.9 pp, medium confidence |
| Gemini 2.0 Flash-Lite | 12.5% | 9.5% | estimated ± 3.9 pp, medium confidence |
| Gemini 1.5 Flash | 9.6% | 7.4% | estimated ± 3.9 pp, medium confidence |
Fitted on
| Model | SWE-bench Verified | AA-SciCode |
|---|---|---|
| Apodex 1.1 | 77.7% | 45.5% |
| Claude Fable 5 | 95.0% | 61.0% |
| Claude Opus 4.8 | 88.6% | 54.4% |
| Claude Opus 5 | 96.0% | 56.4% |
| Claude Sonnet 5 | 85.2% | 54.3% |
| DeepSeek V3 | 42.0% | 35.8% |
| DeepSeek V4 Flash 0731 | 79.0% | 50.3% |
| DeepSeek V4 Pro 0813 | 80.6% | 51.0% |
| Gemini 2.5 Pro | 63.8% | 46.3% |
| GPT-OSS 120B | 62.4% | 34.0% |
| GPT-OSS 20B | 60.7% | 38.9% |
| Granite 4.2 30B | 57.0% | 37.8% |
| Granite 4.2 8B | 47.7% | 31.5% |
| Hy3 Preview | 74.4% | 48.6% |
| Inkling | 77.6% | 47.0% |
| Inkling-Small | 80.2% | 49.7% |
| K-EXAONE 2.0 | 68.2% | 42.0% |
| Kimi K2.6 | 80.2% | 51.5% |
| MiniCPM5-2B | 46.4% | 26.3% |
| MiniMax M3 | 80.5% | 47.1% |
| Mistral Medium 3.5 128B | 77.6% | 40.2% |
| Muse Glimmer 30B | 76.0% | 44.9% |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 52.8% | 32.1% |
| Nemotron 3.5 Super VL 120B A12B | 71.1% | 47.8% |
| Nemotron 3.5 Super VL 120B A12B NVFP4 | 71.1% | 47.8% |
| Nemotron 3 Ultra | 71.9% | 40.3% |
| Qwen3.5-122B-A10B | 72.0% | 39.7% |
| Qwen3.5 397B | 76.2% | 44.8% |
| Qwen3.6-27B | 77.2% | 42.8% |
| Qwen3.6-35B-A3B | 73.4% | 36.6% |
| Qwen3.7 Max | 80.4% | 49.5% |
| Qwen3.7 Plus | 77.7% | 46.1% |
| Beam 1.0 501B-A23B | 81.6% | 48.5% |
| Solar Pro 4 | 70.6% | 48.0% |