LATIDIA · Investigación

SpireBench: Evaluando el juego estratégico de Long-Horizon de modelos genéticos con Slay the Spire

arXiv: 2610.12596v1Tipo de anuncio: nuevo Resumen: Los modelos genéticos de lenguaje grande ahora abordan tareas complejas, pero el razonamiento estratégico en horizontes largos está poco estudiado: la mayoría de los puntos de referencia agénticos son cortos y episódicos, y por

La noticia

arXiv:2610.12596v1 Announce Type: new Abstract: Agentic large language models now tackle complex tasks, yet strategic reasoning over long horizons is under-studied: most agentic benchmarks are short and episodic, and performance on long tasks is hard to dissect into stored knowledge, reasoning and luck. We introduce SpireBench, in which models play the complete commercial deck-builder Slay the Spire (StS), where early decisions decide long-term success. Across 20 large language models (LLMs) at the game's highest difficulty, frontier models beat heuristic baselines and match the average human player but remain far behind experts. This gap widens as the run lengthens: the strongest models win the first act as often as experts, then