27B Local Model From StartLux Ranks Second in Official Chinese Agent Benchmark, Matching Larger Rivals on Key Tasks

A 27-billion-parameter local model from Shanghai-based StartLux scored 39.25 and ranked second in a China Academy of Information and Communications Technology agent-capability test, according to multiple technology outlets citing the academy’s report. The model outperformed larger DeepSeek variants on several multi-tool tasks.（本文作者为 Chelsea_Sun，钛媒体经授权发布）NextFin News — A compact local model has rec

原文：钛媒体