12 10-second Vidu Q4 Preview clips vs Sume at 720p

Vidu Q4 Preview, released on 2026-10-07, advertises launch pricing "from $0.014 a second", so 12 10-second clips come to $1.68 at that starting rate. The cheapest 720p model in Sume's video catalog for the same length is MiniMax H3 at $9.00. Vidu's number is a starting rate, and its release does not say which resolution it applies to, so the two columns below are a floor against a list price, not a like-for-like quote.

Vidu's side comes from its launch release and Unite.AI's summary, read 2026-10-10. Sume's side comes from the Video Router docs and the API pricing page, computed from the same estimators that bill a job.

What do 12 10-second clips cost at 720p?

Sume prices are the provider list price times 1.25, which is what the ledger charges. Vidu is shown at its starting rate only; Unite.AI reports the model offers 3 to 16 second clips at 540p, 720p, 1080p, 2K and 4K, so a 10-second clip is in range.

How do I run the Sume side?

Submit to POST /v1/videos with the catalog id, then poll the polling_url. This example runs the cheapest row above and needs a Sume API key in SUME_API_KEY.

import os
import time

import requests

H = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}
job = requests.post(
    "https://api.sume.com/v1/videos",
    headers=H,
    json={
        "model": "minimax-h3",
        "prompt": "A paper boat drifts down a rain-soaked street at dusk",
        "duration": 10,
        "resolution": "768p",
        "aspect_ratio": "16:9",
    },
).json()
while True:
    s = requests.get(job["polling_url"], headers=H).json()
    if s["status"] in ("completed", "failed"):
        break
    time.sleep(10)
print(s["status"], s.get("unsigned_urls"))

What this comparison leaves out

Read these before treating either number as a quote.

• Vidu's starting rate is the floor of its rate card; a 1080p, 2K or 4K clip may cost more per second.
• Sume does not list Vidu Q4 Preview in its video catalog at the time of reading, so there is no Sume price for it.
• Quality, motion and audio behavior differ by model and are not scored here.