Veo 3.1 1080p needs 8 seconds: the short-clip trap and a 3 s option

You cannot make a 4 second 1080p clip with Veo 3.1. The Veo guide says an 8 second duration is required for 1080p, for 4K, and when you use reference images. At Standard rates, one 8 second 1080p clip costs $3.20. On Sume, gemini-omni-flash-1.1 makes a 3 second 1080p clip for $0.5625 before rounding.

What the 8 second rule costs on Veo

Prices are Google's paid-tier per-second rates from the Gemini API pricing page.

Why it hurts short clips

The trap is that a social cut that needs a 3 or 4 second 1080p shot still pays for 8 seconds, then you trim the rest. The same rule applies if you add reference images at any resolution, because the guide ties reference images to 8 seconds too.

The Sume alternative

The Sume Video Router table lists gemini-omni-flash-1.1 at 360p, 720p, 1080p and 4K for 3 to 10 seconds. The Sume 1080p rate is $0.1875 per second (fal list $0.15 times 1.25). Veo 3.1 itself is not in the Sume catalog, so this is a different model, not a drop-in switch.

Paying for seconds you throw away

Say you need a 4 second 1080p shot. Veo has to render 8 seconds and you keep half. Omni on Sume renders the 4 seconds you asked for. The table divides what you pay by the seconds you keep (4 s), using the rows above.

Check before you switch

Sume says the billable amount is rounded up to cents, so the 3 second row bills about $0.57. Omni on Sume accepts image and video input_references too (up to 10 images and 3 clips of 3 seconds or less), and the Sume docs do not tie them to an 8 second minimum.

• Omni on Sume is 16:9 or 9:16 only, like Veo.
• Audio is always on for Omni; generate_audio: false returns 400.
• Verify a 3 second clip fits the shot: Omni clips are 3 to 10 seconds, so a 2 second shot still needs a trim.