Calibration

QwenClawBench → MCP Atlas

Name: benchgap: LLM benchmark scores, measured and estimated
Creator: benchgap

MCP Atlas is estimated from QwenClawBench with a offset logistic curve fitted on 8 models measured on both: y = 0.4237 + (0.7481 − 0.4237) / (1 + exp(−200.00·(x − 0.5797))), R² = 0.67, cross-validated error 12.3 pp. It is used for 1 estimate.

Estimated model	QwenClawBench	MCP Atlas	Source
Qwen 3.6 Max (preview)	59.0%	71.1%	estimated ± 12.3 pp, low confidence