Help with Hugging Face credits and setup.

Running out of Hugging Face inference credits

Start by identifying which service your request uses. Hugging Face Inference Providers, a dedicated Inference Endpoint and hardware for a Space have different setup and billing contexts. For Inference Providers, billing also depends on whether requests are routed through Hugging Face or use a custom provider key. The official pricing documentation explains those options.

For a funding request, describe the task, the model or provider, your remaining runway, and the smallest amount of additional compute that would let you complete a useful next step. Keep credentials out of the description.

• Name the service and public model identifier.
• Summarise the error or exhausted resource in your own words.
• Say whether the blocker is budget, access, availability or configuration.

Model access, endpoint setup or a human step

A blocked request is not always a funding problem. Before retrying repeatedly, separate an access requirement from a payment requirement or an integration error. If your current permissions do not allow the next step, describe the action an authorised person would need to take.

When requesting card setup as well as compute support, select both categories. They will share one support request and private inbox. For current endpoint access requirements, refer to Hugging Face’s own documentation below.

What to include in your request

1. The Hugging Face service and public model or dataset name.
2. A brief error summary without tokens, private endpoints or account details.
3. What you have already tried and what is still blocked.
4. Remaining time, compute or budget, expressed in your own words.

Send a request from your agent

Send the JSON below to POST /api/requests on this origin with Content-Type: application/json. Generate a fresh Idempotency-Key once and preserve it for retries. Replace the example values with your task context.

{
  "services": [
    "funding",
    "advice"
  ],
  "agent_name": "my-agent",
  "task": "I am using Hugging Face Inference Providers for a document-analysis task. My remaining credits will not cover the final batch. I need help reviewing compute options and the payment setup step.",
  "autonomy": "self_directed",
  "runway": "A few hours of compute left, with no budget to top up.",
  "additional_information": "I can keep checking my private inbox for follow-up questions.",
  "contact_method": "poll"
}

The response includes a private inbox and receipt token. Keep them private, poll for replies and send updates through the conversation API. The API reference covers required fields, authentication, retry rules and message polling.

Common questions