Calibration
Toolathlon-Verified → MCP Atlas
MCP Atlas is estimated from Toolathlon-Verified with a Michaelis–Menten curve fitted on 11 models measured on both: y = 1.4626·x / (0.57698 + x), R² = 0.61, cross-validated error 7.8 pp. It is used for 2 estimates.
| Estimated model | Toolathlon-Verified | MCP Atlas | Source |
|---|---|---|---|
| Claude Sonnet 5.5 | 77.8% | 84.0% | estimated ± 7.8 pp, medium confidence |
| Laguna S 2.1 | 49.7% | 67.7% | estimated ± 7.8 pp, medium confidence |
Fitted on
| Model | Toolathlon-Verified | MCP Atlas |
|---|---|---|
| Claude Opus 5 | 80.6% | 85.8% |
| DeepSeek V4 Flash 0731 | 70.3% | 69.0% |
| DeepSeek V4 Pro 0813 | 74.1% | 73.6% |
| Hy4 preview | 74.1% | 83.7% |
| Inkling-Small | 54.4% | 79.6% |
| Kimi K3 | 73.2% | 84.2% |
| Muse Spark Open | 72.8% | 89.2% |
| Ornith-1.5-35B-A3B | 48.7% | 70.2% |
| Ornith-1.5-397B | 71.2% | 80.0% |
| Ornith-1.5-9B | 41.2% | 54.2% |
| Step 5 Preview | 74.1% | 85.6% |