**["European Sovereign AI" Dropping the Act? Mistral Directly Invites Zhipu GLM-5.2 In]** According to monitoring by Beating, Mistral, which has consistently championed "European Sovereign AI," has now directly integrated a Chinese model into its platform. Mistral announced it would begin hosting third-party open models, with the first being GLM-5.2 from Zhipu's Z.ai. The model runs directly on Mistral's infrastructure, enjoying the same regional control and service guarantees as Mistral's own models. Previously, Mistral had been mocked by the community alongside DeepSeek. After the release of Mistral Large 3, researchers dissected the model configuration and found that it essentially adopted the same architecture as DeepSeek V3/V3.1, with adjustments mainly in expert configurations and other details. Now, Mistral has openly brought a Chinese model into its "sovereign AI" system. Some in the community are joking that European sovereign AI teams won't even need to repackage Chinese models anymore—they can simply provide European servers for them. Jokes aside, Mistral has actually redefined "sovereign AI": models don't necessarily need to be trained by European companies. The key is that enterprises can choose their own models, control where the models run, where the data resides, and who holds the computational power. Thus, a Chinese model running on European infrastructure can still qualify as "European Sovereign AI" under Mistral's framework. [Original Link]