auray-client

The Auray API from Python. One dependency, httpx; Python 3.10+.

pip install auray-client

import uuid

from auray_client import Client

auray = Client()  # reads AURAY_API_KEY

result = auray.subscribe(
    "auray-ai/krea-2-turbo/text-to-image",
    input={"prompt": "a brass orrery on a walnut desk, morning light", "aspect_ratio": "16:9", "resolution": 1024},
    idempotency_key=str(uuid.uuid4()),  # one per submit: a fixed key hands back the first job
    on_queue_update=lambda s: print(s["status"]),
)
assets = auray.platform.jobs.assets(result["request_id"])
print(assets["assets"][0]["url"])

• run(address, input=..., wait_seconds=..., timeout=...) — the synchronous host; a 504 run_timed_out is followed by the job it names (queue_request_id, never the trace in request_id), never resubmitted. wait_seconds is how long the host holds the connection, timeout the whole call's deadline; running out raises subscribe_timed_out carrying the job.
• subscribe(address, input=..., idempotency_key=..., on_queue_update=...) — submit, follow, fetch the result, waiting as long as the server's give_up_after_seconds says unless timeout says otherwise.
• queue.submit / status / result / cancel / stream_status — the request lifecycle, fal's shape.
• storage.upload(address=..., content_type=..., data=...) and upload_file(path, address=...) — three calls, a signed URL, and the field it goes in.
• platform.me / usage / plans / catalogue / status / changelog / jobs / models / webhooks / chat — the account's side.
• embeddings.create(model=..., input=...) — vectors from the OpenAI-compatible route.
• AsyncClient — the same surface, awaited: async with AsyncClient() as auray: ...

Every refusal raises AurayError with status, code, error_type, the API's message, docs and request_id, and the body's own retry and charged: a refusal is resent only when that advice says the identical request may be answered (same_key, after_retry_after) and no charge stands.

Documentation: https://auray.ai/docs/clients/python