New 'Queen' AI plays chess like a Grandmaster and explains its moves

Researchers have developed "Queen," a 4-billion-parameter language model capable of playing chess at a Grandmaster level and explaining its moves. The model integrates a specialized chess encoder with an instruction-tuned language model, trained through an iterative distillation process that refines explanations over multiple cycles. This approach significantly boosted Queen's Elo rating from 1782 to 2697, surpassing other frontier models in both playing strength and puzzle accuracy, while its explanations were found to be coherent and comparable to GPT-5.6 "Sol". The framework is suggested as a transferable method for applying language models to other domains with available expert encoders. AI

IMPACT Demonstrates a new method for integrating expert knowledge into LLMs, potentially advancing AI capabilities in complex, rule-based domains.

RANK_REASON Publication of a research paper detailing a new AI model with novel capabilities. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• alphaXiv
• arXiv
• Bellman update
• CatalyzeX
• DagsHub
• Gotit.pub
• GPT-5.6 "Sol"
• Grandmaster
• Hugging Face
• Queen
• ScienceCast

• model release
• paper
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →