Omni Flash 1.1 is billed in tokens: 5,792 per 720p second, in dollars

One second of 720p Gemini Omni Flash 1.1 video is 5,792 output tokens, and at $17.50 per million video output tokens that is $0.10136, which Google rounds to about $0.10 per second. Sume lists the same model as gemini-omni-flash-1.1 at $0.125 per 720p second, which is $0.10 times the 1.25 house multiplier.

This post does the token arithmetic once so you can budget an Omni job without opening a calculator, and shows where Sume's per-second rate comes from. The Google figures are from its Gemini API pricing page, read 2026-10-09, where the model is described as generally available on the paid tier.

From tokens to a clip

Google states billing is based on total output token consumption at 5,792 tokens per second of 720p video, with video output at $17.50 per million tokens and text output at $9.00 per million. Multiply seconds by 5,792, then by 17.50 divided by 1,000,000.

Where Sume differs

Sume bills per output second, not per token, so there is no token count to track. The catalog row accepts 3 to 10 seconds in 16:9 or 9:16, with native synced audio always on, and the API rejects generate_audio: false. Sume's other published rates for the same row are:

• 360p: $0.0375 per second
• 720p: $0.125 per second
• 1080p: $0.1875 per second
• 4K: $0.375 per second

Reading the gap

For a 10-second 720p clip Google's token math gives $1.0136 and Sume charges $1.250, a difference of $0.2364. That difference is the Sume multiplier applied to a rounded $0.10, not a separate fee. Google's text input is $1.50 per million tokens for text, image, video and audio inputs; Sume's catalog price is quoted for the output second at default options.

If you send reference images or an edit source, check the Video Router docs for the allowed inputs: up to 10 reference images, up to 3 reference videos of at most 3 seconds each, and a video_url edit mode. Pull the live per-second SKUs from GET /v1/videos/models before you commit to a budget.

Budget examples

The following figures use the same list prices as the tables above.

• A creator posting 30 ten-second 720p clips a month spends 30 x $1.25 = $37.50 on Sume's Omni row. At Google's token math the same month is 30 x $1.0136 = $30.41.
• A 3-second 720p clip is 17,376 tokens, or $0.3041 at Google's rate; Sume bills 3 x $0.125 = $0.375 and reserves 38 cents once rounded up to a whole cent.
• Because Sume prices by the second and not by token, the number does not move with the prompt length or any thinking tokens that Google counts in its output price.
• Always compare like with like: Google's $17.50 figure is for video output tokens, while its text output price is $9.00 per million.