Skip to main content

Overview

The Model endpoint is the primary entry point for generating content with the Higgsfield API. Each model on the Higgsfield platform has a unique model_id (e.g., higgsfield-ai/soul/standard). Base API URL: https://platform.higgsfield.ai The Higgsfield API uses an asynchronous request-response pattern. When you submit a generation request, it enters a queue and processes in the background. This approach offers several advantages:
  • Monitor task status without maintaining open connections
  • Cancel requests that haven’t started processing
  • Avoid resource-intensive concurrent connections

API Endpoints

Queue Management

Parameters

  • model_id: The unique identifier for the model (e.g., higgsfield-ai/soul/standard)
  • request_id: A unique UUID assigned to your request upon submission

Usage Examples

Submitting a Generation Request

Response Format

Queued Request

Completed Request

Request Statuses

Canceling a Request

You can cancel a request only while it remains in the queued status. Once processing begins, cancellation is no longer possible.
If cancellation was successful, you will get a 202 Accepted response status code. Otherwise, response status code will be 400 Bad Request.

Best Practices

  • Poll the status endpoint periodically to check request progress
  • Or use webhooks to get generation result by HTTP
  • Store the request_id to retrieve results later
  • Handle different status codes appropriately in your application