Enregistré dans:
Détails bibliographiques
Auteurs principaux: Kolasani, Sai, Saplin, Maxim, Crispino, Nicholas, Montgomery, Kyle, Davis, Jared Quincy, Zaharia, Matei, Wang, Chi, Wang, Chenguang
Format: Preprint
Publié: 2025
Sujets:
Accès en ligne:https://arxiv.org/abs/2512.01992
Tags: Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
_version_ 1866915647254953984
author Kolasani, Sai
Saplin, Maxim
Crispino, Nicholas
Montgomery, Kyle
Davis, Jared Quincy
Zaharia, Matei
Wang, Chi
Wang, Chenguang
author_facet Kolasani, Sai
Saplin, Maxim
Crispino, Nicholas
Montgomery, Kyle
Davis, Jared Quincy
Zaharia, Matei
Wang, Chi
Wang, Chenguang
contents We introduce LLM CHESS, an evaluation framework designed to probe the generalization of reasoning and instruction-following abilities in large language models (LLMs) through extended agentic interaction in the domain of chess. We rank over 50 open and closed source models by playing against a random opponent using a range of behavioral metrics, including win and loss rates, move quality, move legality, hallucinated actions, and game duration. For a subset of top reasoning models, we derive an Elo estimate by playing against a chess engine with variably configured skill, which allows for comparisons between models in an easily understandable way. Despite the simplicity of the instruction-following task and the weakness of the opponent, many state-of-the-art models struggle to complete games or achieve consistent wins. Similar to other benchmarks on complex reasoning tasks, our experiments reveal a clear separation between reasoning and non-reasoning models. However, unlike existing static benchmarks, the stochastic and dynamic nature of LLM CHESS uniquely reduces overfitting and memorization while preventing benchmark saturation, proving difficult even for top reasoning models. To support future work on evaluating reasoning and instruction-following in LLMs, we release our experimental framework, a public leaderboard, and a dataset of associated games.
format Preprint
id arxiv_https___arxiv_org_abs_2512_01992
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle LLM CHESS: Benchmarking Reasoning and Instruction-Following in LLMs through Chess
Kolasani, Sai
Saplin, Maxim
Crispino, Nicholas
Montgomery, Kyle
Davis, Jared Quincy
Zaharia, Matei
Wang, Chi
Wang, Chenguang
Artificial Intelligence
Computation and Language
We introduce LLM CHESS, an evaluation framework designed to probe the generalization of reasoning and instruction-following abilities in large language models (LLMs) through extended agentic interaction in the domain of chess. We rank over 50 open and closed source models by playing against a random opponent using a range of behavioral metrics, including win and loss rates, move quality, move legality, hallucinated actions, and game duration. For a subset of top reasoning models, we derive an Elo estimate by playing against a chess engine with variably configured skill, which allows for comparisons between models in an easily understandable way. Despite the simplicity of the instruction-following task and the weakness of the opponent, many state-of-the-art models struggle to complete games or achieve consistent wins. Similar to other benchmarks on complex reasoning tasks, our experiments reveal a clear separation between reasoning and non-reasoning models. However, unlike existing static benchmarks, the stochastic and dynamic nature of LLM CHESS uniquely reduces overfitting and memorization while preventing benchmark saturation, proving difficult even for top reasoning models. To support future work on evaluating reasoning and instruction-following in LLMs, we release our experimental framework, a public leaderboard, and a dataset of associated games.
title LLM CHESS: Benchmarking Reasoning and Instruction-Following in LLMs through Chess
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2512.01992