LATIDIA · Robótica

Mismas piezas, diferentes servidores: un punto de referencia de Tetris para los agentes de IA según lo servido

arXiv:2603.02348v2 Tipo de anuncio: replace-cross Resumen: Un agente cumple con un modelo como servido: a través de un punto final con una tarjeta de precio, una caché compartida y otros inquilinos, o en cualquier hardware que ejecute un modelo autohospedado. Banco

La noticia

arXiv:2603.02348v2 Announce Type: replace-cross Abstract: An agent meets a model as served: through an endpoint with a price card, a shared cache and other tenants, or on whatever hardware a self-hosted model runs. Benchmarks rank the weights. We introduce a Tetris benchmark that measures what agents get from models as served: every move is scored against an oracle, and every agent receives the same pieces. In five pre-specified experiments with nine open-weight models on one serverless provider, plus an open decision model self-hosted on a CPU, we find that price and size do not predict decision quality; that resending history costs almost nothing when cached input is free, would cost