Calibration

DeepPlanning → MCP-Tasks

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

MCP-Tasks is estimated from DeepPlanning with a Hill curve fitted on 5 models measured on both: y = 0.0000 + (0.7515 − 0.0000)·x^2.89 / (0.09027^2.89 + x^2.89), R² = 1.00, cross-validated error 1.3 pp. It is used for 0 estimates.