15 menu special clips of 5s: Wan 3.0 vs Omni Flash cost

15 menu special clips of 5 seconds each is 75 seconds of video, which costs $4.69 on Sume with Wan 3.0 at 480p, $9.38 at 720p and $18.75 at 1080p. Gemini Omni Flash 1.1 at 720p is $9.38 and at 1080p $14.06.

This is the cost of one batch of short menu special clips, priced before you submit anything.

What does the whole job cost?

Both models bill per second of output video, by resolution. Wan 3.0's rate card has 480p, 720p and 1080p rows; Gemini Omni Flash 1.1 has 360p, 720p, 1080p and 4K rows. The table prices this job's total seconds at the rows that matter, as read from Sume's rate card on 2026-10-10.

Which limits apply?

The Sume video docs list the limits by model (read 2026-10-10): wan-3.0 accepts 2 to 30 seconds per clip, and gemini-omni-flash-1.1 accepts 3 to 10 seconds at 16:9 or 9:16 with native synced audio. Your 5-second clips are inside both ranges.

• Ask GET /v1/videos/models for the exact resolutions, aspect ratios and durations your account accepts before you submit.
• Add a retry allowance: 20 percent more seconds on $9.38 is $11.25 at Wan 3.0 720p. The 20 percent is an assumption.
• Draft at 480p ($0.0625 per second), then rerun only the approved clips at a higher tier.

A request that runs

Video generation is asynchronous: POST /v1/videos returns a job with a polling_url; poll it until status is completed, then read unsigned_urls. Send an Idempotency-Key so a retry returns the original job instead of a second paid one.

import os
import time
import requests

H = {"Authorization": f"Bearer {os.environ['SUME_API_KEY']}"}

def main():
    job = requests.post(
        "https://api.sume.com/v1/videos",
        headers={**H, "Idempotency-Key": "15-menu-special-clips-5-seconds-wan-3-0-vs-omni-"},
        json={"model": "wan-3.0", "prompt": "a menu special scene, steady camera",
              "duration": 5, "resolution": "480p", "aspect_ratio": "9:16"},
        timeout=60,
    ).json()
    while job["status"] in ("pending", "in_progress"):
        time.sleep(30)
        job = requests.get(job["polling_url"], headers=H, timeout=60).json()
    print(job["status"], job.get("unsigned_urls"))

main()

Where to go next

See list video models for the catalog call, every AI model on Sume for the models, and how Sume pricing works for the billing rules.