Calibration
AA-SciCode → FrontierSWE v2
FrontierSWE v2 is estimated from AA-SciCode with a linear curve fitted on 17 models measured on both: y = 3.0836·x + -1.4255, R² = 0.49, cross-validated error 14.6 pp. It is used for 48 estimates.
| Estimated model | AA-SciCode | FrontierSWE v2 | Source |
|---|---|---|---|
| A.X K2 | 41.0% | 0.0% | estimated ± 14.6 pp, low confidence |
| Apodex 1.1 | 45.5% | 0.0% | estimated ± 14.6 pp, low confidence |
| Apodex 1.1 Mini | 45.5% | 0.0% | estimated ± 14.6 pp, low confidence |
| Celeris-1 | 21.6% | 0.0% | estimated ± 14.6 pp, low confidence |
| Claude Haiku 5.5 | 55.0% | 27.1% | estimated ± 14.6 pp, low confidence |
| Command A+ | 38.5% | 0.0% | estimated ± 14.6 pp, low confidence |
| DeepSeek V3 | 35.8% | 0.0% | estimated ± 14.6 pp, low confidence |
| DeepSeek V3 0324 | 39.0% | 0.0% | estimated ± 14.6 pp, low confidence |
| DeepSeek V4.1 Flash | 51.9% | 17.5% | estimated ± 14.6 pp, low confidence |
| Gemma 3 27B | 23.3% | 0.0% | estimated ± 14.6 pp, low confidence |
| Gemma 4 26B A4B | 40.0% | 0.0% | estimated ± 14.6 pp, low confidence |
| Gemma 4 31B | 45.5% | 0.0% | estimated ± 14.6 pp, low confidence |
| Gemma 4 E4B | 24.4% | 0.0% | estimated ± 14.6 pp, low confidence |
| GPT-6.1 Sol | 54.2% | 24.6% | estimated ± 14.6 pp, low confidence |
| GPT-6 Luna | 54.6% | 25.8% | estimated ± 14.6 pp, low confidence |
| GPT-6 Sol | 57.6% | 35.1% | estimated ± 14.6 pp, low confidence |
| GPT-OSS 120B | 34.0% | 0.0% | estimated ± 14.6 pp, low confidence |
| GPT-OSS 20B | 38.9% | 0.0% | estimated ± 14.6 pp, low confidence |
| Granite 4.2 3B | 25.3% | 0.0% | estimated ± 14.6 pp, low confidence |
| Hy3 | 48.6% | 7.3% | estimated ± 14.6 pp, low confidence |
| Hy3 Preview | 48.6% | 7.3% | estimated ± 14.6 pp, low confidence |
| K-EXAONE 2.0 | 42.0% | 0.0% | estimated ± 14.6 pp, low confidence |
| LFM2.5-2.6B | 14.4% | 0.0% | estimated ± 14.6 pp, low confidence |
| Ling 3.0 Flash FP8 | 42.0% | 0.0% | estimated ± 14.6 pp, low confidence |
| Ling 3.0 Flash VL | 44.2% | 0.0% | estimated ± 14.6 pp, low confidence |
| Ling 3.0 Tiny | 24.2% | 0.0% | estimated ± 14.6 pp, low confidence |
| Ling 3.1 Flash | 54.1% | 24.3% | estimated ± 14.6 pp, low confidence |
| Llama 4 Maverick | 31.7% | 0.0% | estimated ± 14.6 pp, low confidence |
| Llama 4 Scout | 21.3% | 0.0% | estimated ± 14.6 pp, low confidence |
| Mercury 2.5 | 38.5% | 0.0% | estimated ± 14.6 pp, low confidence |
| MiMo-V2.6-Flash | 51.3% | 15.6% | estimated ± 14.6 pp, low confidence |
| MiMo-V2.6-Pro | 60.9% | 45.2% | estimated ± 14.6 pp, low confidence |
| Mistral Large 3 | 36.6% | 0.0% | estimated ± 14.6 pp, low confidence |
| Mistral Large 4 | 54.2% | 24.6% | estimated ± 14.6 pp, low confidence |
| Mistral Small 4 | 38.8% | 0.0% | estimated ± 14.6 pp, low confidence |
| Mistral Small 4 (Reasoning) | 38.8% | 0.0% | estimated ± 14.6 pp, low confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 32.1% | 0.0% | estimated ± 14.6 pp, low confidence |
| Nemotron 3 Nano 30B | 30.6% | 0.0% | estimated ± 14.6 pp, low confidence |
| Nemotron 3 Super 100B | 36.2% | 0.0% | estimated ± 14.6 pp, low confidence |
| North Mini Code | 38.8% | 0.0% | estimated ± 14.6 pp, low confidence |
| Quasar 438B | 48.1% | 5.8% | estimated ± 14.6 pp, low confidence |
| Qwen3.5-122B-A10B | 39.7% | 0.0% | estimated ± 14.6 pp, low confidence |
| Qwen3.8 Max Preview | 52.1% | 18.1% | estimated ± 14.6 pp, low confidence |
| Solar Pro 3 | 25.5% | 0.0% | estimated ± 14.6 pp, low confidence |
| Solar Pro 4 | 44.6% | 0.0% | estimated ± 14.6 pp, low confidence |
| Step 5 Preview | 58.9% | 39.1% | estimated ± 14.6 pp, low confidence |
| Trinity-Large-Preview | 40.6% | 0.0% | estimated ± 14.6 pp, low confidence |
| Trinity-Large-Thinking | 40.6% | 0.0% | estimated ± 14.6 pp, low confidence |