Calibration
AA-SciCode → FrontierSWE v2
FrontierSWE v2 is estimated from AA-SciCode with a Hill curve fitted on 20 models measured on both: y = 0.0000 + (1.2000 − 0.0000)·x^6.00 / (0.66441^6.00 + x^6.00), R² = 0.49, cross-validated error 14.2 pp. It is used for 33 estimates.
| Estimated model | AA-SciCode | FrontierSWE v2 | Source |
|---|---|---|---|
| Step 5 Preview | 58.9% | 39.2% | estimated ± 14.2 pp, low confidence |
| GPT-6.1 Terra | 55.1% | 29.5% | estimated ± 14.2 pp, low confidence |
| Qwen3.8 Max Preview | 52.1% | 22.6% | estimated ± 14.2 pp, low confidence |
| MiMo-V2.6-Flash | 51.3% | 21.0% | estimated ± 14.2 pp, low confidence |
| Hy3 | 48.6% | 15.9% | estimated ± 14.2 pp, low confidence |
| Hy3 Preview | 48.6% | 15.9% | estimated ± 14.2 pp, low confidence |
| Quasar 438B | 48.1% | 15.1% | estimated ± 14.2 pp, low confidence |
| Solar Pro 4 | 48.0% | 14.9% | estimated ± 14.2 pp, low confidence |
| Gemini 2.5 Pro | 46.3% | 12.3% | estimated ± 14.2 pp, low confidence |
| Apodex 1.1 | 45.5% | 11.2% | estimated ± 14.2 pp, low confidence |
| Apodex 1.1 Mini | 45.5% | 11.2% | estimated ± 14.2 pp, low confidence |
| Gemma 4 31B | 45.5% | 11.2% | estimated ± 14.2 pp, low confidence |
| Ling 3.0 Flash VL | 44.2% | 9.6% | estimated ± 14.2 pp, low confidence |
| K-EXAONE 2.0 | 42.0% | 7.2% | estimated ± 14.2 pp, low confidence |
| A.X K2 | 41.0% | 6.3% | estimated ± 14.2 pp, low confidence |
| Trinity-Large-Preview | 40.6% | 5.9% | estimated ± 14.2 pp, low confidence |
| Trinity-Large-Thinking | 40.6% | 5.9% | estimated ± 14.2 pp, low confidence |
| Gemma 4 26B A4B | 40.0% | 5.5% | estimated ± 14.2 pp, low confidence |
| Nemotron 3.5 Bolt 74B-A9B | 39.7% | 5.2% | estimated ± 14.2 pp, low confidence |
| Qwen3.5-122B-A10B | 39.7% | 5.2% | estimated ± 14.2 pp, low confidence |
| DeepSeek V3 0324 | 39.0% | 4.7% | estimated ± 14.2 pp, low confidence |
| GPT-OSS 20B | 38.9% | 4.6% | estimated ± 14.2 pp, low confidence |
| Mistral Small 4 | 38.8% | 4.6% | estimated ± 14.2 pp, low confidence |
| North Mini Code | 38.8% | 4.6% | estimated ± 14.2 pp, low confidence |
| Command A+ | 38.5% | 4.4% | estimated ± 14.2 pp, low confidence |
| Mercury 2.5 | 38.5% | 4.4% | estimated ± 14.2 pp, low confidence |
| Mistral Large 3 | 36.6% | 3.3% | estimated ± 14.2 pp, low confidence |
| DeepSeek V3 | 35.8% | 2.9% | estimated ± 14.2 pp, low confidence |
| GPT-OSS 120B | 34.0% | 2.1% | estimated ± 14.2 pp, low confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 32.1% | 1.5% | estimated ± 14.2 pp, low confidence |
| Llama 4 Maverick | 31.7% | 1.4% | estimated ± 14.2 pp, low confidence |
| Mistral Small 3.1 | 27.8% | 0.6% | estimated ± 14.2 pp, low confidence |
| MiniCPM5-2B | 26.3% | 0.5% | estimated ± 14.2 pp, low confidence |
Fitted on
| Model | AA-SciCode | FrontierSWE v2 |
|---|---|---|
| Claude Fable 5 | 61.0% | 47.0% |
| Claude Fable 5.1 | 63.1% | 56.3% |
| Claude Fable 5.2 | 67.7% | 69.9% |
| Claude Haiku 5.5 | 55.0% | 43.8% |
| Claude Opus 5 | 56.4% | 52.0% |
| Claude Opus 5.5 | 66.9% | 62.3% |
| Claude Sonnet 5.5 | 61.0% | 61.9% |
| Gemini 3.7 Flash | 57.2% | 20.3% |
| Gemini 3.8 Flash | 56.6% | 19.6% |
| Gemini 4 Argon | 61.8% | 55.0% |
| GLM-5.3 | 59.0% | 30.2% |
| GPT-5.6 Sol | 57.1% | 32.2% |
| GPT-6 Astra | 56.5% | 65.5% |
| Grok 4.6 | 56.5% | 25.3% |
| Grok 4.7 | 57.4% | 29.5% |
| Inkling | 47.0% | 4.1% |
| Inkling 2 Small | 52.3% | 25.1% |
| Kimi K3 | 59.5% | 25.9% |
| Muse Spark 1.2 | 57.4% | 12.0% |
| Muse Spark 1.3 | 58.8% | 25.8% |