Calibration
AA-AnalystAgent → Toolathlon-Verified
Toolathlon-Verified is estimated from AA-AnalystAgent with a linear curve fitted on 6 models measured on both: y = 0.2257·x + 0.6564, R² = 0.70, cross-validated error 1.9 pp. It is used for 4 estimates.
| Estimated model | AA-AnalystAgent | Toolathlon-Verified | Source |
|---|---|---|---|
| Claude Fable 5 | 48.8% | 76.7% | estimated ± 1.9 pp, medium confidence |
| GPT-6.1 Sol | 50.0% | 76.9% | estimated ± 1.9 pp, medium confidence |
| Inkling | 23.8% | 71.0% | estimated ± 1.9 pp, low confidence |
| Nemotron 3 Ultra | 6.3% | 67.1% | estimated ± 1.9 pp, low confidence |