AI Radar

Noteworthy AI conversations and writing, summarized from publisher notes.

NVIDIA Acquires Hugging Face for $13B; Z.ai Launches GLM-5.3-Flash Model

· Latent Space

NVIDIA has acquired Hugging Face for $13B, nearly double its initial $7B offer. Concurrently, Z.ai launched its GLM-5.3-Flash model, previously known as "Ox Alpha." This model features 320B total and 18B active parameters, a 1M-token context window, and is natively multimodal, released under an MIT License. Z.ai claims it runs entirely on Chinese AI chips and outperforms GLM-5.2 while matching Claude Opus 4.8 in coding. Independent evaluations by Artificial Analysis show an Intelligence Index score of 57 and a cost of $0.09 per task, positioning it as a highly cost-effective option. The model has seen rapid adoption, with Cline reporting it as its fastest-growing model, driving 11% of traffic in under a week, and infrastructure providers like CoreWeave and Baseten quickly offering support.

AI Radar wrote this summary from the publisher’s notes with AI, not from a transcript. How it works.