Calibration
SWE Multilingual → MILU
MILU is estimated from SWE Multilingual with a Hill curve fitted on 5 models measured on both: y = 0.8880 + (1.2000 − 0.8880)·x^6.00 / (1.26794^6.00 + x^6.00), R² = 0.62, cross-validated error 0.9 pp. It is used for 4 estimates.
| Estimated model | SWE Multilingual | MILU | Source |
|---|---|---|---|
| Claude Mythos 5 | 92.2% | 92.8% | estimated ± 0.9 pp, medium confidence |
| Ling 3.1 Flash | 88.5% | 92.0% | estimated ± 0.9 pp, low confidence |
| Claude Opus 4.8 | 84.4% | 91.3% | estimated ± 0.9 pp, low confidence |
| Hy4 preview | 82.9% | 91.1% | estimated ± 0.9 pp, low confidence |
Fitted on
| Model | SWE Multilingual | MILU |
|---|---|---|
| Claude Fable 5.1 | 89.1% | 93.0% |
| Claude Fable 5.2 | 96.6% | 94.1% |
| Claude Opus 5 | 89.5% | 92.1% |
| Claude Opus 5.5 | 93.9% | 93.1% |
| Claude Sonnet 5.5 | 90.3% | 91.6% |