Queen: A 4B Chess LM Hits Grandmaster Level (2697 Elo) While Explaining Its Moves
Adithya Bhaskar · hf · 2026-10-05
Researchers introduce Queen, a 4B-parameter chess-language model that plays at Grandmaster level (2697 Elo) while explaining its moves and plans in natural language.
Method:
- Architecture: a "silent expert" chess encoder is wired to an instruction-tuned LM via cross-attention, with a QA curriculum teaching the LM to extract chess concepts from the encoder's representations.
- Iterative distillation: a natural-language analog of the Bellman update — the model analyzes positions after its top candidate moves, consolidates them into an explanation of the current position, and distills that back into itself.
Results: over 7 iterations the model gains 900+ Elo (1782 → 2697), surpassing all frontier models in both playing strength and puzzle accuracy despite three orders of magnitude fewer parameters; its explanations approach GPT-5.6-Sol (high) in coherence.
The authors frame this as a general recipe for any domain with a silent expert encoder: games, robotics, and computer use.
More from Research
- PCA vs EGA on LLM embeddings: EGA recovers 6-dim structure 97.6–100%, PCA nearly 0% — GolinoHudson · 2026-10-06
- Stanford studies find AI benchmarks may not measure what they claim, with billions riding on scores — StanfordHAI · 2026-10-06
- Reverse engineering Neuralink's new decoder architecture from a sparse blog post — melnykowycz · 2026-10-06
- Harvard-MIT paper: "plan ahead" prompts make agents play worse; interface design matters more — dair_ai · 2026-10-06
- First Human–AI Interaction Conference HAIC 2027 announced for June in Washington, DC — jasonwuishere · 2026-10-06
- Gym-Anything gets oral presentation at COLM 2026 Lifelong Agents Workshop — wellecks · 2026-10-06