Calibration
τ³-Bench Banking → AA-AnalystAgent
AA-AnalystAgent is estimated from τ³-Bench Banking with a linear curve fitted on 7 models measured on both: y = 1.2775·x + -0.1076, R² = 0.89, cross-validated error 8.1 pp. It is used for 4 estimates.
| Estimated model | τ³-Bench Banking | AA-AnalystAgent | Source |
|---|---|---|---|
| Qwen3.8 Max | 51.3% | 54.8% | estimated ± 8.1 pp, low confidence |
| Qwen3.8 2.4T A95B | 49.1% | 52.0% | estimated ± 8.1 pp, low confidence |
| Qwen3.8 27B (medium) | 47.4% | 49.8% | estimated ± 8.1 pp, low confidence |
| Muse Spark 1.3 (xhigh) | 47.2% | 49.5% | estimated ± 8.1 pp, low confidence |