Our paper SokoBench: Evaluating Long-Horizon Planning and Reasoning in Large Language Models was accepted at TMLR.