Coke is an important raw material for blast furnace ironmaking, and its quality is an important factor affecting the quality of molten iron and the smooth operation of blast furnace. In order to solve the problems of difficult inspection, hysteresis, and large prediction errors in coke quality, a coke prediction model based on gradient boosting decision tree algorithm is proposed. Combined with expert experience and correlation analysis, the influence of mixed coal quality on coke quality is studied. Finally, the mixed coal quality parameters are used to predict the ash content, sulfur content, M10 and M40 of coke quality parameters. The model is evaluated based on the historical production data of a coking plant. The experimental results show that the coke quality prediction model based on the gradient boosting decision tree has less error and higher accuracy than the linear regression model, random forest model, and decision tree model. It can provide a certain theoretical basisfor coal blending and coking of the coking plant.