Flow Veo 3.1 Quality has no ingredients: references on Sume Omni
In Flow, Veo 3.1 Quality takes text and frames only, while Fast and Lite take ingredients. Sume's reference-to-video runs on Omni with up to 10 images.
In Google Flow, Veo 3.1 Quality supports text to video and frames to video but not Ingredients/References, whereas Veo 3.1 Fast and Lite do (read 2026-10-02). If you need reference images and the top Veo tier, Flow makes you choose. On Sume you pick gemini-omni-flash-1.1 and send reference image URLs; Sume does not list Veo.
The Flow matrix
Flow's help page gives supported features per model (Flow Help, read 2026-10-02). Durations for the Veo models are 4, 6 and 8 seconds; Omni Flash 1.1 lists 4, 6, 8 and 10.
| Flow model | Frames to video | Ingredients or references |
|---|---|---|
| Veo 3.1 Lite | First, first and last | Yes (8s limit) |
| Veo 3.1 Fast | First, first and last | Yes |
| Veo 3.1 Quality | First, first and last | No |
| Gemini Omni Flash 1.1 | First, first and last | Yes |
What the API adds for Omni
The Gemini API page for Omni Flash has a reference_to_video task for subject reference images, and allows video references of at most 3 clips, up to 3 seconds each. It does not support multi-video prompting (Gemini API docs, read 2026-10-02).
The Sume request
On Sume, gemini-omni-flash-1.1 takes reference_image_urls (up to 10) and reference_video_urls (up to 3, each up to 3 seconds) in the Video Router, addressed in the prompt as <IMAGE_REF_0> and <VIDEO_REF_0> by list order (Video Router). Clips run 3 to 10 seconds, 16:9 or 9:16, with native audio.
On POST /v1/videos the same idea is input_references. If you send both frame_images and input_references, frame_images wins and the call is treated as image to video (Video generation). Pick one per request.
What you give up
Sume has no Veo 3.1 tier, so there is no Quality-tier look to choose. If you specifically want Veo's output, that stays a Flow or Gemini API decision. If you want references with one billed API, Omni covers it, and the price is the catalog's per-second rate by resolution times Sume's margin, readable from the models endpoint.
Sources
Related posts
More in Models
- FLUX.2 takes 8 references by API, 10 in Playground: Sume's range
Black Forest Labs says FLUX.2 pro, max and flex take up to 8 references by API (10 in its Playground), klein 4. Sume's input_references range is read per model.
- flux-2-pro-preview vs flux-2-pro: which id does Sume send?
BFL has flux-2-pro-preview (latest) and flux-2-pro (fixed snapshot). Sume lists black-forest-labs/flux.2-pro and does not let you pick the BFL endpoint.
- gemini-omni-1.1-flash vs gemini-omni-flash-1.1: which id goes where
Google writes the model id gemini-omni-1.1-flash; Sume's catalog id is gemini-omni-flash-1.1. A mapping table, the old preview id, and how to avoid a typo.
- Gemini Omni audio reference: unsupported; Sume models that take audio
Google says Omni's API doesn't accept uploaded audio references. Sume's Omni row has no reference_audio_urls; Seedance 2.x, Wan 3.0 and MiniMax H3 honor audio.
Written by Sume