Wan 3.0 vs Gemini Omni Flash: same $0.125/s at 720p, pick by limits

At 720p, Wan 3.0 and Gemini Omni Flash 1.1 both cost $0.125 per second on Sume, so price will not choose for you. Wan runs 2 to 30 seconds and takes audio references. Omni runs 3 to 10 seconds, reaches 4K, edits existing clips, and has native synced audio that you cannot turn off. At 1080p Omni is cheaper.

Price per second

Both rows are list x 1.25. Wan 3.0 lists $0.05, $0.10 and $0.20 per second at 480p, 720p and 1080p (fal, 2026-08-24, as recorded in the repo). Omni lists $0.03, $0.10, $0.15 and $0.30 at 360p, 720p, 1080p and 4K (fal, 2026-08-28).

Limits side by side

The Video Router docs and model cards give the envelopes.

Choosing

Use Wan 3.0 for anything over 10 seconds, for clips under 3 seconds, for square or 4:3 frames, and when you want to feed it a voiceover or track as a reference. A 30 s clip at 720p is $3.75.

Use Omni for 1080p work (a 10 s clip is $1.88 against Wan's $2.50), for 4K, for editing a finished clip with video_url, and for the default Auto route. Both take image references; Omni needs <IMAGE_REF_0> style tags in the prompt to address them.

Same prompt, two models

Because the request shapes line up on /v1/videos, you can A/B the same prompt by changing only model. Keep the duration inside the overlap (3 to 10 s) and the ratio at 16:9 or 9:16. Because both rows are the same price at 720p, the cost of the test is the clip length times $0.125, and the comparison is about picture and sound only. See the reference limits before you attach inputs.

for m in wan-3.0 gemini-omni-flash-1.1; do
  curl -s -X POST https://api.sume.com/v1/videos \
    -H "Authorization: Bearer $SUME_API_KEY" \
    -H "Content-Type: application/json" \
    -d "{\"model\":\"$m\",\"prompt\":\"A street musician plays violin at dusk\",\"duration\":6,\"resolution\":\"720p\",\"aspect_ratio\":\"16:9\"}"
  echo
done

A budget view

For 20 clips of 8 s at 720p, both models cost $20.00. At 1080p, Omni is $30.00 and Wan is $40.00. At the Wan-only 30 s length a single 1080p clip is $7.50, and a Wan 480p draft of the same length is $1.88. Use those figures to decide whether a long-shot plan or a short-shot plan is cheaper for your story.

The Videos page lists Wan 3.0 as a picker option and reaches Omni through Auto, so you can compare both without leaving the page.