Calibration
Claw-Eval → MCP Atlas
MCP Atlas is estimated from Claw-Eval with a Hill curve fitted on 16 models measured on both: y = 0.0000 + (0.8902 − 0.0000)·x^6.00 / (0.56593^6.00 + x^6.00), R² = 0.58, cross-validated error 11.2 pp. It is used for 11 estimates.
| Estimated model | Claw-Eval | MCP Atlas | Source |
|---|---|---|---|
| DeepSeek V3.2 | 40.2% | 10.1% | estimated ± 11.2 pp, low confidence |
| GLM-5-Turbo | 55.8% | 42.6% | estimated ± 11.2 pp, low confidence |
| GLM-5V-Turbo | 53.8% | 37.8% | estimated ± 11.2 pp, low confidence |
| K-EXAONE 2.0 | 77.7% | 77.5% | estimated ± 11.2 pp, low confidence |
| LLaDA2.2-mini | 57.2% | 45.8% | estimated ± 11.2 pp, low confidence |
| MiMo-V2.5 | 62.3% | 57.0% | estimated ± 11.2 pp, low confidence |
| MiMo-V2-Omni | 45.2% | 18.3% | estimated ± 11.2 pp, low confidence |
| MiMo-V2-Pro | 57.8% | 47.3% | estimated ± 11.2 pp, low confidence |
| Ornith-1.0-35B | 69.8% | 69.3% | estimated ± 11.2 pp, low confidence |
| Ornith-1.0-397B | 77.1% | 77.0% | estimated ± 11.2 pp, low confidence |
| Ornith-1.0-9B | 63.1% | 58.5% | estimated ± 11.2 pp, low confidence |