bfl-api

Send a POST request to a model endpoint like /v1/flux-2-pro with your prompt and dimensions, authenticated via the x-key header. The response returns a polling_url which you GET repeatedly until the status is Ready, then download the result image.

The standard limit is 24 concurrent in-flight requests. On HTTP 429, read the Retry-After header and retry with exponential backoff, or use a semaphore-based client to cap concurrency proactively.

Use polling for scripts, CLI tools, and simple integrations since it works everywhere. Use webhooks for production server-to-server workloads where you need immediate notification and want to avoid repeated polling requests.

Result URLs from the BFL API expire 10 minutes after generation completes. Download the image immediately when the polling status becomes Ready instead of storing or caching the URL.

Yes, all FLUX.2 models accept input_image parameters for editing, with URLs preferred over base64. Klein models support up to 4 reference images, while pro, max, and flex support up to 8.

When you set a webhook_secret, BFL signs payloads with HMAC-SHA256 in the X-BFL-Signature header. Compute the expected signature over the raw request body and compare it using a constant-time comparison function.