Releasesby Glyphhuggingface_hub [v1.32.0] Shared blob store, sandbox security hardening and faster imports
📂 [Cache] Shared blob store: deduplicate Xet files across repos The cache now deduplicates Xet files across repos. A Xet file downloaded through hf_xet is stored once at <CACHE_DIR>/blobs/<prefix>/<xet_hash> and every repo that needs it gets a relative symlink instead of a download: no bytes are transferred and no extra space is used, even across different...
Also shipped Sep 17, 2026Hugging Face in September 2026 →
You can add unit tests to pipelines to validate your SQL transformation logic against mock datasets
You can add unit tests to pipelines to validate your SQL transformation logic against mock datasets. Unit tests for pipelines are generally available (GA).
Metastore-level ABAC policies are in Beta
Metastore-level ABAC policies are in Beta in Unity Catalog. You can now attach an ABAC policy (row filter, column mask, GRANT, or DENY) at the metastore level, so a single policy applies across every catalog in the metastore, including catalogs created later.
Moonshot AI Kimi K2.7 is scheduled for retirement
Moonshot AI Kimi K2.7 will be retired on October 30, 2026. Kimi K3 is the recommended replacement. See Deprecated and retired models for migration guidance.
HTML generation in Snowflake CoWork (Preview)
HTML generation is now available in preview as part of document generation in Snowflake CoWork. You can ask Snowflake CoWork to generate an HTML file from your analysis, the same way you generate a PDF or PowerPoint. Generation requires the Cortex Agent code execution tool on the associated agent.
Base environments are now supported on classic compute (Beta)
You can now use base environments to manage Python dependencies on classic compute. You can use Databricks-provided environments or custom workspace-level environments. See Manage dependencies using environments .
Snowpark Container Services backup instance types (General availability)
Backup instance types for Snowpark Container Services compute pools are now generally available and are no longer in Preview .
More from Hugging Face HubEvery Hugging Face Hub release →
[v2.0.0] The httpx2 release
See MIGRATION GUIDE in docs. 💔 Breaking Change: HTTP stack moves to httpx2 See migration plan in #4802 . huggingface_hub v2.0 replaces its HTTP dependency with httpx2 : clients and transport exceptions now come from httpx2 / httpcore2 instead of httpx / httpcore .
[v1.33.0] Better UX on large uploads and simpler skills installs
📤 See progress while large Xet uploads validate Very large Xet uploads used to sit silently between "Uploading" and "Committing" while shards were being finalized, which could take a long time with no feedback.
Preview LeRobot Episodes
LeRobot datasets now open with an episode preview on the dataset page. Every camera plays on one shared transport, so you can scrub through an episode and see each angle at the same moment. Dataset listings also show robot type, episode count, and a preview of the first episode.
Hugging Face is now on Google Cloud Marketplace
Organizations can now subscribe to Hugging Face with their Google Cloud account. Usage for Inference Providers, Inference Endpoints, Spaces, Jobs, storage, and Team or Enterprise seats will appear on their Google Cloud invoice.
[v1.31.0] Custom labels for Sandboxes, More resilient downloads and more
🏷️ Custom labels for Sandboxes Dedicated sandboxes now accept custom labels, attached to the underlying Job. This is useful for cost attribution, bookkeeping, or finding and reconnecting to the sandboxes created by a given controller run. Labels use the same -l / --label KEY=VALUE syntax as hf jobs run .
Live resource usage on Jobs
The job page has a new Resources panel next to the logs. While your job runs, it shows live CPU, RAM, GPU and VRAM usage, updated every couple of seconds.
Hugging Face Hub 2.1.1
[CLI] Rename 'hf spaces zero-gpu-quota' to 'hf spaces zero-gpu quota'- #5055
[v2.1.0] Better Jobs tooling, ZeroGPU quota tracking, and revamped Model Catalog
📊 Jobs: retries, rescheduling, and port exposure Hugging Face Jobs get a batch of quality-of-life improvements this release. Jobs can now retry automatically on failure with the new attempts parameter, and you can start a fresh Job from an existing Job's saved spec (including secrets and hardware settings) with rerun_job / hf jobs rerun.