Beijing AI filings reach 205 as city approves 2 new genAI services

Beijing quietly crossing 205 registered generative AI services is less about two new models and more about the maturing of China’s AI governance stack. The city is now running a standing registry that forces providers to disclose model names, filing numbers and content‑labeling practices, turning what used to be a wild‑west release process into something more like a regulated utilities regime.([finance.sina.com.cn](https://finance.sina.com.cn/tech/roll/2025-12-24/doc-inhcwtcp4521732.shtml))

For builders chasing AGI‑class systems within China, this is the environment they now operate in. Compliance with labeling and filing rules becomes a non‑negotiable part of deployment, which will shape which architectures, data sources and business models are viable at scale. At the same time, a large, visible registry lowers the barrier for downstream apps: product teams can choose from a catalog of approved models without guessing which ones pass regulatory muster.

Globally, Beijing’s model registry is an early real‑world implementation of “AI system registries” that Western policymakers often talk about but haven’t yet fully rolled out. It hints at a future where high‑capability models can’t be quietly shipped; they have to live on government lists with transparency around their intended use. That may not slow core research much, but it will influence where and how AGI‑aspiring models are productized.