Tag
Google's Gemma4-12B model weights on HuggingFace have been silently updated; the reason for the update is unclear but may involve fixes.
Meta served a legal notice to the Heretic Project over derivatives of its Llama AI models, prompting the project to remove the weights and announce plans to diversify infrastructure with an official Codeberg mirror.
Hugging Face launches Storage Buckets, a purpose-built storage solution for AI teams offering per-TB pricing, built-in CDN, and Xet deduplication for model weights, datasets, and checkpoints.
OpenAI submitted a comment to the NTIA outlining their historical approach to model weight distribution, from GPT-2's staged release to GPT-3's API-first strategy, while discussing the trade-offs between open-source model releases and controlled deployment through commercial products.