Hugging Face CLI now called 'hf' with cleaner commands and new Jobs feature for Pro users

Hugging Face has officially rebranded its command-line interface (CLI) from huggingface-cli to hf, marking a major update aimed at improving usability and streamlining workflows for developers and researchers. The change, announced as a “quality-of-life improvement,” reflects the team’s effort to simplify the CLI’s command structure and align it with modern practices. The decision to rename the tool stems from user feedback and the evolving complexity of its features. Over time, the CLI’s commands—such as upload, download, cache management, and repo management—became increasingly fragmented, making navigation less intuitive. By adopting the hf name, the team seeks to create a more ergonomic and discoverable interface. The new structure follows a predictable pattern: hf , which organizes commands by their purpose and improves clarity. For example, tasks like authentication, caching, and repository management are now grouped under dedicated subcommands, such as hf auth or hf repo. To begin using the updated CLI, users must install the latest version of huggingface_hub via pip: bash pip install -U huggingface_hub After restarting the terminal, running hf version confirms the installation, displaying the current version number. The root-level commands now prioritize frequently used actions like upload and download, while subcommands like auth, cache, and repo provide more specialized functionality. For deeper guidance on any command, users can append --help to explore options. A key shift in the migration process is the restructuring of authentication commands. Previously, users would run huggingface-cli login or huggingface-cli whoami, but these have been reorganized under hf auth. For instance: bash hf auth login hf auth whoami hf auth logout The legacy huggingface-cli remains functional but will display warnings directing users to the new syntax. This transition aims to reduce confusion while maintaining backward compatibility. The CLI also introduces a new feature: hf jobs, a service enabling users to execute scripts or Docker images on Hugging Face’s infrastructure. Pro users and teams can leverage this tool to run workloads on customizable hardware, with billing based on usage time. For example, launching a script on an A10G GPU involves: bash hf jobs run --flavor=a10g-small ubuntu nvidia-smi The command structure mirrors Docker’s familiar syntax, offering subcommands like inspect, logs, ps, run, and cancel to manage tasks efficiently. Additionally, uv allows direct execution of Python scripts with inline dependencies, expanding the CLI’s capabilities for developers. Hugging Face emphasizes that the hf CLI is designed to adapt to future innovations, ensuring a scalable foundation for new tools. The company also highlights its commitment to maintaining the legacy CLI during the transition period, though users are encouraged to adopt the updated version for improved performance and features. This update underscores Hugging Face’s focus on enhancing developer experience, addressing pain points in AI workflows, and reinforcing its role as a central platform for model training and deployment. For detailed instructions, users can refer to the official documentation or explore the hf jobs guide to unlock advanced capabilities.