Current Google Flow video models
The model selector is not only a quality choice. It changes which inputs, durations and edit actions are available.
| Capability | Veo 3.1 Lite | Veo 3.1 Fast | Veo 3.1 Quality | Gemini Omni Flash |
| Text to video | 4s, 6s, 8s | 4s, 6s, 8s | 4s, 6s, 8s | 4s, 6s, 8s, 10s |
| First frame | Supported | Supported | Supported | Supported |
| First + last frame | Supported | Supported | Supported | Listed as coming soon |
| Ingredients/references | 8s | 8s | Not supported | 4s–10s with advanced references |
| Video-to-video editing | Not supported | Not supported | Not supported | Up to 10-second selected segment |
| Extend | Used to extend supported Veo 3.1 8s clips | Not the extension model | Not the extension model | Listed as coming soon |
Google’s current help table should be treated as the source of truth when a feature in the interface differs.
When to choose Veo 3.1 Lite
Lite currently has the lowest credit cost and is the model Google specifies for extending eligible Veo 3.1 eight-second clips. It is a practical choice for iteration volume and extension workflows.
When to choose Veo 3.1 Fast
Fast uses more credits than Lite but fewer than Quality for non-Ultra users. It supports the common text, frame and ingredient workflows documented for Veo 3.1.
When to choose Veo 3.1 Quality
Quality currently costs 100 credits per generation and supports eight-second output. Use it when the intended workflow justifies the higher allowance cost, not as an automatic default for every exploratory prompt.
When to choose Gemini Omni Flash
Omni Flash currently adds:
10-second generation.
Editing of uploaded and generated video.
Advanced character, avatar and supported audio references.
Custom voice styling from a preset voice.
Some Omni Flash features have country restrictions.
Current image models
Google currently lists Nano Banana Pro, Nano Banana 2 Lite and Nano Banana 2 for image generation and editing inside Flow. The default can depend on the subscription.
A model-selection checklist
Choose in this order:
Required input: text, frames, ingredients or uploaded video.
Required duration.
Whether you need editing or extension.
Country availability.
Credit cost per output and expected retries.
Then confirm the active model and cost in the prompt-box settings before generating.
Sources and verification
Sources support product facts; editorial judgment remains FlowTool's own. Links were last checked on the date shown above.
Frequently asked questions
Which Google Flow model can generate 10-second clips?
Google currently lists 10-second generation for Gemini Omni Flash. Veo 3.1 models are listed with 4-, 6- and 8-second lengths.
Which model edits uploaded videos in Google Flow?
Google currently documents uploaded and generated video editing for Gemini Omni Flash.
Which model extends Google Flow videos?
Google says eligible eight-second Veo 3.1 clips can be extended and that Veo 3.1 Lite is used for the extension.
Visible FAQ is provided for readers. FlowTool does not promise FAQ rich results.
Continue with FlowTool
Build a clearer Google Flow workflow.
Use the beginner workflow, then choose a prompt template or compare alternatives based on your actual constraints.
Open the beginner tutorial
