Calibration
MCP Atlas → Claw-Eval
Claw-Eval is estimated from MCP Atlas with a linear curve fitted on 16 models measured on both: y = 0.3266·x + 0.4507, R² = 0.53, cross-validated error 5.7 pp. It is used for 13 estimates.
| Estimated model | MCP Atlas | Claw-Eval | Source |
|---|---|---|---|
| Claude Opus 5 | 85.8% | 73.1% | estimated ± 5.7 pp, low confidence |
| DeepSeek V4 Flash 0731 | 69.0% | 67.6% | estimated ± 5.7 pp, medium confidence |
| DeepSeek V4 Pro 0813 | 73.6% | 69.1% | estimated ± 5.7 pp, medium confidence |
| GPT-5.4 mini | 57.7% | 63.9% | estimated ± 5.7 pp, medium confidence |
| GPT-5.4 nano | 56.1% | 63.4% | estimated ± 5.7 pp, medium confidence |
| Inkling | 74.1% | 69.3% | estimated ± 5.7 pp, medium confidence |
| Inkling-Small | 79.6% | 71.1% | estimated ± 5.7 pp, medium confidence |
| Kimi K2.7 Code | 76.0% | 69.9% | estimated ± 5.7 pp, medium confidence |
| LongCat-Flash-Lite-Sparse | 45.6% | 60.0% | estimated ± 5.7 pp, medium confidence |
| Muse Glimmer 30B | 75.5% | 69.7% | estimated ± 5.7 pp, medium confidence |
| Beam | 78.7% | 70.8% | estimated ± 5.7 pp, medium confidence |
| Solar Open 2 | 58.2% | 64.1% | estimated ± 5.7 pp, medium confidence |
| Solar Pro 4 | 61.4% | 65.1% | estimated ± 5.7 pp, medium confidence |