Jingle mascot clip: 6 s on Wan 3.0 with an audio_url reference, $0.75

A 6-second Wan 3.0 clip driven by an audio reference costs $0.375 at 480p, $0.75 at 720p and $1.50 at 1080p on Sume. Wan 3.0 is one of the models that accepts audio references, and the price is the per-second rate whether you send one or not (read 2026-10-09).

Where an audio reference goes

The docs list audio_url among supported_input_references for the Seedance 2.x models, Wan 3.0 and the MiniMax H3 family. For Wan 3.0 the catalog allows up to 5 audio references. You attach the file in input_references with a type of audio_url.

Request

The audio URL must be public HTTPS. Add a visual reference image in the same array to steer the look.

import os, time, requests

API = "https://api.sume.com/v1/videos"
H = {"Authorization": "Bearer " + os.environ["SUME_API_KEY"]}
body = {"model": "wan-3.0", "prompt": "A cartoon mascot bounces to the beat of the jingle in a bright studio", "duration": 6, "resolution": "720p", "aspect_ratio": "1:1", "input_references": [{"type": "audio_url", "audio_url": {"url": "https://example.com/jingle.mp3"}}, {"type": "image_url", "image_url": {"url": "https://example.com/mascot.png"}}]}
job = requests.post(API, headers={**H, "Idempotency-Key": "jingle-mascot-001"}, json=body, timeout=60).json()
deadline = time.time() + 1800
while time.time() < deadline:
    time.sleep(30)
    job = requests.get(job["polling_url"], headers=H, timeout=60).json()
    if job["status"] == "completed":
        print(job["unsigned_urls"][0], job["usage"]["cost"])
        break
    if job["status"] in ("failed", "cancelled"):
        print(job["status"], job.get("error"))
        break

Ten product jingles

Ten 6-second jingle clips at 720p cost 10 x $0.75 = $7.50. At 480p the batch is $3.75, at 1080p $15.00.

A request with an audio reference follows the same flow as any other video job: submit, poll, download from unsigned_urls[0]. usage.cost on the completed response shows the billable amount.

If you leave the audio reference out and set generate_audio to true, the model can make its own soundtrack. The price tool takes no audio argument for Wan 3.0, which reflects that the estimate depends on resolution and seconds.

Gotchas

• Reference audio is guidance, not a guaranteed lip-sync or exact sync. Check the result before you lay it over a final edit.
• Keep the jingle at or under the clip length. The Wan 3.0 catalog entry caps reference audio at 5 files and 15 seconds in total; read the live limits from GET /v1/video-router/models.
• Not every model takes audio: Gemini Omni Flash 1.1 takes video references but not audio.
• If the job fails, read the error field, confirm the URLs are reachable over HTTPS, and retry with the same idempotency key.