HuggingFace

Hugging Face is the collaboration platform for the machine learning community, built for exploring and sharing models, datasets, and applications across many modalities. You can browse 2M+ models and discover 500k+ datasets, plus explore Spaces and community-built AI apps. It’s also a great place to build and share a public portfolio of your ML work.

Hugging Face includes access to a broad open-source ML tooling ecosystem, including Transformers, Diffusers, Datasets, and Safetensors, along with additional libraries for training, deploying, and running models (such as PEFT and Accelerate). You can also use the Hub for hosting and collaborating on unlimited public models, datasets, and applications.

If you’re working as an organization or need enterprise-grade controls, Hugging Face offers Team & Enterprise plans starting at $20/user/month. These plans include features like Single Sign-On (SSO), audit logs, resource groups, priority support, and private dataset viewing, along with Inference Providers. You can also deploy using Inference Endpoints, or update Spaces applications to run on GPU in a few clicks.

For inference, Hugging Face provides access to 45,000+ models from leading AI providers through a single unified API, with no service fees. For compute, Inference Endpoints are priced starting at $0.60/hour for GPU and are billed by usage. Spaces can run with free CPUs or be upgraded to GPU/accelerators, with persistent storage costing between $5/month and $100/month. Public repositories are free, while PRO and Enterprise plans add additional capabilities.

What can I do on Hugging Face?

Browse and collaborate on AI models, datasets, and applications (including Spaces). You can also build using the Hugging Face open-source ML tooling stack.

Is it free to host public models, datasets, and apps on the Hub?

Unlimited public repositories are available at no cost. Paid plans add additional capabilities beyond public hosting.

How do Inference Providers work on Hugging Face?

Inference Providers give access to 45,000+ models from leading providers through a single unified API, with no service fees.

What is the pricing model for Inference Endpoints?

Inference Endpoints are usage-based and billed per hour based on the CPU instance type and resources you use. GPU pricing starts at $0.60/hour.

How does Spaces GPU and storage pricing work?

Spaces can run on free CPUs or be upgraded to GPU/accelerators, with prices varying by hardware. Persistent storage costs between $5/month and $100/month.

What’s included in Hugging Face Team & Enterprise plans?

Plans start at $20/user/month and include features like Single Sign-On (SSO), audit logs, resource groups, private dataset viewing, regions, priority support, and Inference Providers.