AI BriefingHugging Face Integrates Serverless Inference
Key point
Hugging Face has directly integrated major serverless inference providers into the Hub and SDK.
Details
Hugging Face has directly integrated four major serverless inference providers—fal, Replicate, Sambanova, and Together AI—into model pages on the Hub. This makes it easier for developers to explore and utilize serverless inference across various models.
Key Features:
- User-customized settings: Users can register each provider's API key directly in settings, and arrange widgets in their preferred provider order.
- Two calling modes: Both a 'Custom Key' method using the provider's own API key, and a 'Routed by HF' method that bills through a Hugging Face account without needing a separate key, are supported.
- SDK integration: Through the Python
huggingface_hub(v0.28.0 or later) and JS SDK, a specific provider's infrastructure can be called with just a few lines of code.
With this update, an environment has been established for instantly testing and prototyping the latest models, such as DeepSeek-R1, across various infrastructure environments.
This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.
Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.