huggingface_hub has reached version 1.0 after five years of development, marking a milestone in its maturity as a Python package that powers 200,000 dependent libraries and provides core functionality for accessing over 2 million public models, 0.5 million public datasets, and 1 million public Spaces. The release introduces breaking changes designed to support the next decade of open machine learning, driven by a global community of almost 300 contributors and millions of users. Major changes include a migration to httpx as the backend library, a completely redesigned hf CLI that replaces the deprecated huggingface-cli and features a Typer-based interface with an expanded feature set, and full adoption of hf_xet for file transfers, replacing the legacy hf_transfer. The library has evolved from a simple Git wrapper for transformers checkpoints to a comprehensive platform that now supports Spaces, inference endpoints, and a Jobs API, with a growing ecosystem of third‑party integrations. The migration to Xet has moved 77PB+ across 6,000,000 repositories, improving upload and download efficiency without user intervention. While most libraries remain backward compatible, transformers v4 requires huggingface_hub v0.x and the upcoming transformers v5 will require v1.x. The upgrade brings native HTTP/2 support, true thread safety, and a unified API for synchronous and asynchronous operations, positioning huggingface_hub to scale with the explosive growth of AI while maintaining reliability for millions of developers.
huggingface_hub v1.0: Five Years of Building the Foundation of Open Machine Learning
huggingface_hub has released version 1.0 after five years of development, marking its maturity as a Python package that powers 200,000 dependent libraries and provides core functionality for accessing over 2 million public models, 0.5 million public datasets, and 1 million public Spaces. The release introduces breaking changes, including a migration to httpx, a redesigned hf CLI, and full adoption of hf_xet for file transfers, while maintaining backward compatibility for most libraries. The library’s evolution has expanded from a simple Git wrapper to a comprehensive platform supporting model hosting, Spaces, inference endpoints, and a Jobs API, with a growing ecosystem of third‑party integrations.
Why it matters
The upgrade positions huggingface_hub to scale with AI growth, offering modern HTTP/2 support and unified sync/async APIs, which benefits developers and the broader machine learning ecosystem.