Reference-to-video, 5 seconds: Vidu Q4 $0.1575 discounted vs Sume rows

A 5-second reference-to-video clip costs $0.1575 at Vidu Q4's discounted 540P rate and $1.365 at 4K. Sume's rows in the table run $0.50 to $1.89.

5 min readSume
All posts

The short answer

A 5-second reference-to-video clip on Vidu Q4 Preview costs 5 x $0.0315 = $0.1575 at 540P and 5 x $0.273 = $1.365 at 4K while the 30% discount lasts, and $0.225 and $1.95 at list. Sume does not list Vidu. The Sume rows that accept references include Omni, Seedance 2.x, Wan 3.0 and the MiniMax H3 pair; in the table below they cost between $0.50 and $1.89 for the same 5 seconds.

Clip cost by row

Every Sume row is a valid length at 5 seconds (Omni 3-10 s, Wan 3.0 2-30 s, MiniMax H3 Max 5-15 s, Seedance 2.0 4-15 s). The Vidu figures use the two endpoints named in the release.

Cost of one 5-second reference-to-video clip, as of 2026-10-08 (Vidu: press release, read 2026-10-08; Sume: public catalog, read 2026-10-08)
RowArithmeticCost
Vidu Q4, 540P, 30% off5 x $0.0315$0.1575
Vidu Q4, 540P, list5 x $0.045$0.225
Vidu Q4, 4K, 30% off5 x $0.273$1.365
Vidu Q4, 4K, list5 x $0.39$1.95
Sume gemini-omni-flash-1.1, 720p5 x $0.125$0.625
Sume gemini-omni-flash-1.1, 1080p5 x $0.1875$0.9375
Sume gemini-omni-flash-1.1, 4K5 x $0.375$1.875
Sume wan-3.0, 720p5 x $0.125$0.625
Sume minimax-h3-max, 768p5 x $0.10$0.50
Sume seedance-2, 720p5 x $0.378$1.89

Reading the gaps

The discounted 540P cell is the cheapest in the table and the only one under 20 cents. At 4K, discounted Vidu is $1.365 and Omni is $1.875, so Vidu is 27% lower. At list, Vidu 4K is $1.95, which is 4% above Omni. The lowest Sume cell is MiniMax H3 Max at 768p, 5 x $0.10 = $0.50.

The cells are not like for like. Vidu 540P is a lower resolution than Sume's 720p and 768p rows, and the Sume rows differ in how closely they follow reference images. The only way to compare that is to send the same references to each and look. Spend the first $2 doing exactly that, with one prompt and one reference set, before you commit a larger batch to any row.

Reference images do not change the Sume price. The docs bill Omni per output second by resolution and the same applies to the other rows. If Vidu charges per reference image the release does not say so.

How to send the references on Sume

Use the flat reference_image_urls field on POST /v1/video-router/generate, or input_references on POST /v1/videos. The two APIs share model ids. Add an Idempotency-Key header on the Video Router so a retry cannot double-bill, and read the model's capabilities from GET /v1/video-router/models first. The Video Router docs show a working request for Omni with video and image references. Keep the references on public HTTPS URLs, because the docs say a reference that cannot be fetched fails the job; check the poll response for the error field before you retry.

Sources

Related posts

More in Pricing

All Pricing posts

Written by Sume