Calibration

QwenClawBench → DeepPlanning

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

DeepPlanning is estimated from QwenClawBench with a inverse Michaelis–Menten curve fitted on 7 models measured on both: y = 0.07547·(x − 0.0000) / (0.0000 + 0.6930 − x), R² = 0.65, cross-validated error 11.8 pp. It is used for 2 estimates.

Estimated model	QwenClawBench	DeepPlanning	Source
Qwen3.6-27B	53.4%	25.3%	estimated ± 11.8 pp, low confidence
Qwen 3.6 Max (preview)	59.0%	43.2%	estimated ± 11.8 pp, low confidence