Changelog

A new Hardware filter on the Models page filters results to models that fit a specific GPU, CPU, or Apple Silicon chip, so you only see what will actually run on your machine. Set the hardware you want from your Hardware settings.

It stacks with the page's other filters and is shareable via URL, so a link like huggingface.co/models?apps=llama.cpp&hardware=apple-m4-max points anyone, even logged-out visitors, at the llama.cpp models that fit an Apple M4 Max.

You can now share your feedback on the Hub directly with the Hugging Face team from the user menu. Whether you want to report a bug, suggest a new feature, or share any other thoughts, your feedback will be carefully reviewed by the team to help shape future improvements.

Enterprise organizations can now create service accounts: dedicated, organization-owned identities for programmatic access such as CI/CD pipelines, automation scripts, and backend integrations. They aren't tied to any individual, so automated workflows keep running even as team members join or leave the organization.

Admins manage them from the organization settings, where access is granted exclusively through fine-grained tokens scoped to what each workflow needs. Tokens can be generated, updated, rotated, or revoked at any time. Service accounts can also be added to resource groups, making it easy to align automated workflows with your organization's existing access structure. Because service accounts don't consume a paid seat, they can be scoped and isolated without impacting member quotas or billing.

Read the documentation to learn more.

You can now publish to Hugging Face repositories from GitHub, GitLab, or custom CI systems without adding secrets, using workflow identity federation.

This mirrors the secret‑less, Trusted Publishing experience available for npm and PyPI. CI jobs can also read from gated repositories that the user has access to. See the documentation

A new Base only toggle on the Models page hides every finetune, adapter, merge, and quantization, leaving just the original base models.

Need the opposite? The Other → Model Tree filter does the reverse, listing only one relation type across the Hub, like every adapter or every quantized model.