Video

Video takes tens of seconds to minutes, so cli2api returns a job ID and the client polls.

POST /v1/videos + GET /v1/videos/{id}

import time, httpx
base = "http://localhost:8080"

# submit
job = httpx.post(f"{base}/v1/videos", json={
    "model": "sora-2",
    "prompt": "a fox jumps over a creek",
    "seconds": "4",
}).json()
job_id = job["id"]            # e.g. "video_a3xq2..."

# poll
while True:
    r = httpx.get(f"{base}/v1/videos/{job_id}").json()
    if r["status"] in ("completed", "failed"):
        break
    time.sleep(5)

print(r.get("videos") or r.get("error"))

status ∈ queued / in_progress / completed / failed.

Supported models

Use the listed name verbatim — t2v / i2v are distinct entries.

Common fields

prompt, negative_prompt, image, first_frame, last_frame, video, reference_images, images, aspect_ratio, resolution, size, duration, seconds, seed, mode, sound, generate_audio, multi_shot, multi_prompt, shot_type, elements, keep_audio. Anything under extra: {...} is forwarded verbatim to the upstream body.