• Optional accessToken: string
The access token to use. Without it, you’ll get rate-limited quickly.
Can be created for free in hf.co/settings/token
You can also pass an external Inference provider’s key if you intend to call a compatible provider like Sambanova, Together, Replicate…
• Optional endpointUrl: string
The URL of the endpoint to use. If not specified, will call huggingface.co/api/tasks to get the default endpoint for the task.
If specified, will use this URL instead of the default one.
• Optional model: string
The HF model to use.
If not specified, will call huggingface.co/api/tasks to get the default model for the task.
/!\ Legacy behavior allows this to be an URL, but this is deprecated and will be removed in the future.
Use the endpointUrl parameter instead.
• Optional provider: "black-forest-labs" | "cerebras" | "cohere" | "fal-ai" | "fireworks-ai" | "hf-inference" | "hyperbolic" | "nebius" | "novita" | "openai" | "replicate" | "sambanova" | "together"
Set an Inference provider to run this model on.
Defaults to the first provider in your user settings that is compatible with this model.