This Zenodo repository provides all the material generated in the research of the paper entiled "Model-based Gradient Search for Permutation Problems" and published in ACM Transactions on Evolutionary Learning and Optimization. Specifically, it is composed of seven zip files each of them with a particular focus. - Instances.zip. This file comprises all the LOP instances used in the paper. Divided into three directories, it groups the IO and xLOLIB instances from one side, and the instances used for the parameter tunning of the algorithms from the other side. - Source Codes.zip. In this file, we will find the source code of the three algorithms, Gradient Search (GS), Natural Evolution Strategy (NES), and Plackett-Luce EDA (PLEDA) implemented in C++. GS and NES algorithms share many components, and for this reason, they are implemented as a part of the same code project. Adaptive versions of the GS and NES are also variants of the same algorithms included in the code project. Finally, a separate file for the PLEDA is provided. The Makefiles in each case are included in the corresponding directories. The remaining 5 files correspond to the different parts of the experimental section. Each of them includes de raw results obtained from the execution of the results (in CSV files), and the scripts to process the results, and generate either figures (the ones that appear in the paper) or to create summary results that are saved again in CSV files. The list is the following:- 1_tunning.zip.- 2_convergence.zip- 3_results.zip.- 4_scalability_time.zip.- 5_adaptive.zip In this sense, the files, as they are numbered, the first one corresponds to the calibration of the parameters made (including the Bayesian statistical assessment), the second file focuses on the convergence analysis of the algorithms, the third one summarises the raw results of the experimentation and performs a Bayesian signed-rank test on the results (by benchmarks and by instances), the fourth file creates the plot that compares the computational cost of the algorithms and analyses the scalability of the algorithms regarding the size of the problems, and finally, the fifth file, compares the adaptive versions of the GS and NES algorithms, by summarizing their results to be shown in tables, and performs Bayesian Signed rank test analysis.