The capabilities, trade-offs and people behind China’s AI models.
A new QbitAI hands-on report describes autonomous game building, Blender modelling and repeated debugging with Kimi K3.1 Pro. The examples are demonstrations reported by the outlet, rather than independent benchmarks.
Ant’s Ling team has announced five Ming-3.0 models spanning speech generation, audio understanding and speech recognition, according to an October 20 report by Phoenix Technology.
An October 20 analysis from Eastmoney Research Center connects recent Chinese model upgrades with enterprise agent adoption and examines the listed-company ecosystem.
National Business Daily’s October 20 analysis puts Chinese models ahead of US models in its OpenRouter-based estimates, while reporting a weekly decline for both groups.
An October 17 report examines Zhipu's GLM-Legal, a specialist model for legal workflows that benchmark provider Vals dates to October 12.
Chinese AI lab DeepSeek released V4.1-Flash-1017 on October 17, claiming stronger benchmark performance, faster task completion, and shorter outputs than its predecessor. The model is live on the company's website, app, and API, with native optimization for Tencent CodeBuddy.
