Replicate runs open models behind one API at https://api.replicate.com/v1. Running a model creates a prediction; add Prefer: wait to get the output in the same response instead of polling.
Copy a token from replicate.com/account/api-tokens and send it as Authorization: Bearer <token>. In the examples it's written {{REPLICATE_API_TOKEN}}.
Full reference: https://replicate.com/docs/reference/http
Generates an image with FLUX Schnell and waits up to 60 seconds for it. output holds the image URL(s).
curl -X POST https://api.replicate.com/v1/models/black-forest-labs/flux-schnell/predictions \
-H "Authorization: Bearer {{REPLICATE_API_TOKEN}}" \
-H "Content-Type: application/json" \
-H "Prefer: wait" \
-d '{"input": {"prompt": "a taco astronaut floating above Earth, digital art"}}'
For predictions still running: poll until status is succeeded or failed.
curl https://api.replicate.com/v1/predictions/{{PREDICTION_ID}} \
-H "Authorization: Bearer {{REPLICATE_API_TOKEN}}"
Description, run count, and the input schema — which parameters the model accepts.
curl https://api.replicate.com/v1/models/black-forest-labs/flux-schnell \
-H "Authorization: Bearer {{REPLICATE_API_TOKEN}}"
The username and account type a token belongs to.
curl https://api.replicate.com/v1/account \
-H "Authorization: Bearer {{REPLICATE_API_TOKEN}}"