Google Flow
Quick Take
Google Flow is where Google puts its generative video and image models to work for people making something longer than a single clip. You create a project, generate or upload shots, keep characters and objects consistent with reference “ingredients,” extend and edit clips, and arrange them into scenes. An agent can handle multi-step jobs, and Flow Tools let you build small custom utilities for jobs you repeat. It is a creative workspace built on models, not a model itself.
How Flow Differs From Gemini, Veo, and AI Studio
Google’s names overlap, so it helps to be precise.
- Veo is a video model. Flow uses Veo 3.1 (in Lite, Fast, and Quality versions) alongside Gemini Omni Flash for video and Nano Banana models for images.
- Gemini is Google’s assistant. It can generate a video in a chat, but it is not built for managing shots, characters, and scenes across a project.
- Google AI Studio is a developer workspace for testing models and getting API keys. Flow is for making media, not building software.
- A traditional video editor gives you a full timeline, precise trims, color, and audio mixing. Flow’s editing is generative: you change shots by prompting, extending, inserting, or removing elements. Many creators generate in Flow and finish in a regular editor.
- Workflow automation platforms such as Zapier or n8n connect apps and move data between them. Flow Tools work only inside Flow, on the media in your project.
Models and What You Can Make
Flow’s help center lists these models as of November 2026:
| Model | What it does in Flow |
|---|---|
| Veo 3.1 Lite | Text, frames, and ingredients to video; the only model that can extend Veo clips |
| Veo 3.1 Fast | Text and frames to video at 4, 6, or 8 seconds |
| Veo 3.1 Quality | Highest-quality Veo clips at 4, 6, or 8 seconds |
| Gemini Omni Flash 1.1 | Text, frames, and ingredients to video up to 10 seconds, plus editing of uploaded and generated videos |
| Nano Banana Pro, 2.1, and 2 Lite | Image generation and editing |
Ingredients are reference images or clips that tell the model what a character, product, or place should look like, and you can call project assets into a prompt by name. Frames let you set the first image, or the first and last, and describe the motion between them. Extend continues a Veo clip. Gemini Omni can edit footage you upload, keep a character’s identity and voice consistent across scenes, and revise a shot through conversation. Scenebuilder arranges clips into a sequence, and camera position and motion controls let you direct shots.
Output can be upscaled to 1080p at no credit cost on paid plans, and to 4K for 50 credits on Ultra. Google watermarks content from its generative models with SynthID, which SynthID Detector can check.
The Flow Agent
Until May 2026, Flow ran one prompt at a time. The Flow Agent, announced on May 19, 2026 and available to all Flow users, takes multi-step requests. It can brainstorm with you (Google’s example is acting as a sounding board for dialogue or plot), generate several variations at once, apply a batch edit across many assets, and organize and rename assets into collections. You stay in control of each step. Agent queries do not use credits, but the generations it starts do.
Flow Tools: Build a Utility Once, Reuse It
Flow Tools, also announced on May 19, 2026, let you describe a small creative utility and have Flow build it. Google’s examples include an image editor, a video resizer, custom shaders, and lo-fi or glitch post-processing effects.
According to Google’s help center, the process works like this:
- Open a project, go to Tools, and choose Create Tool.
- Describe it. Google’s example: “Make an image filter app with a slider for grayscale and hue shift.”
- Flow writes the code and builds an interface with controls. Review it, then select Save.
- To change it later, open the Tool, choose Edit, and describe what to change. You can also rename it, change its description, or set an icon.
Saved Tools appear under My Tools. Favorite makes a Tool available in every project, Pin keeps it handy in the current one, and Remix creates your own copy of someone else’s Tool to change. Media a Tool produces goes into the project you have open.
Sharing is by link. The link shares the version that existed when you created it, not later edits, and anyone with the link can see the Tool’s code, name, and thumbnail. To stop sharing, you have to delete the Tool, so keep anything private out of a Tool you plan to share.
Who can do what. Anyone, including people without a subscription, can use existing Tools. Creating Tools needs a Google AI Plus, Pro, or Ultra subscription, and there is a daily limit on how many you can create. Tools are web-only. Running a Tool that generates media uses credits, and Flow shows a warning when a run might.
Example: a reusable resizing tool
A creator who publishes every video in 16:9 for YouTube, 9:16 for Shorts and Reels, and 1:1 for a feed could ask Flow for “a tool that takes a video from my project and makes 16:9, 9:16, and 1:1 versions, with a slider for where the subject sits in the frame.” After Flow builds it, they test it on one clip, check the generated controls, and ask for changes, perhaps a preview of all three crops side by side. Once it works, they save and favorite it, and it is ready in every later project.
Two cautions. Google lists a video resizer as an example, but it does not document how a generated resizer crops or reframes, and a Tool is code written by a model, so check each output, especially that faces and on-screen text stay in frame. And if the Tool calls a model rather than simply cropping, each run costs credits.
Pricing and Access
Flow runs on credits. Everyone gets 50 a day, which do not roll over. Google AI subscriptions add monthly credits: 200 on Plus, 1,000 on Pro, 10,000 on Ultra ($99.99), and 25,000 on Ultra ($199.99). Costs per generation, from Google’s help center in November 2026:
- Veo 3.1 Lite: 10 credits (5 on Ultra)
- Veo 3.1 Fast: 20 credits (10 on Ultra)
- Veo 3.1 Quality: 100 credits
- Gemini Omni Flash: 7 to 15 credits at 720p, 4 to 7 at 360p, and 40 to edit a video
Subscribers can buy extra credits in most regions, though not in Japan. Without a subscription, video generation may be unavailable during peak hours, roughly 2 PM to 5 PM UTC.
Flow is available in more than 140 countries. There are Android and iOS apps, but the web version is the complete one.
Limitations
- Short clips. Individual generations run 4 to 10 seconds, so longer pieces are built from many clips.
- Credits go quickly on the Quality model; 1,000 monthly Pro credits cover ten Veo 3.1 Quality clips.
- Generative editing, not a full editor. You will likely finish in a traditional editor for precise cuts, mixing, and color.
- Tools are new and partly undocumented. Google does not say exactly how generated Tools handle edge cases, and shared Tools expose their code.
- Mobile is partial. Scenebuilder, camera controls, batch editing, and Tool creation need the web.
- Features vary by plan and region, including some Omni voice features.
Alternatives
- Runway, a video creation platform with its own models, third-party models, node-based Workflows, and App Studio; see Google Flow vs Runway
- Kling AI for video generation at lower cost per clip
- Higgsfield for many video models with cinematic camera controls
- Veo itself, through Gemini or the API, if you only need clips
For the wider category, see the AI video generator glossary entry and our guide to building AI video one shot at a time.
Continue learning
Explore related guides, tools, workflows, and prompts that help you go deeper into this topic.
See how this tool fits into a workflow
Browse step-by-step AI workflows that use ChatGPT, Claude, Gemini, and other tools.
Frequently Asked Questions
What is Google Flow?
Google Flow is an AI creative studio from Google for generating and editing video and images inside projects. It launched at Google I/O 2025 as a filmmaking tool and now uses Veo 3.1, Gemini Omni Flash, and Nano Banana models, with scene building, reference ingredients, an agent for multi-step tasks, and custom Tools. It is available in more than 140 countries at flow.google.com.
Is Google Flow the same as Veo?
No. Veo is a video model. Flow is the application where you use Veo and other Google models: it adds projects, reference ingredients, start and end frames, extending clips, scene building, editing, an agent, and custom Tools around the models.
What are Google Flow Tools?
Flow Tools, announced on May 19, 2026, are small custom utilities you create inside Flow by describing them in plain language, such as an image filter with sliders or a video resizer. Flow writes the code and builds the interface, you review and save it, and you can edit, favorite, pin, remix, or share it later. Building Tools needs a Google AI Plus, Pro, or Ultra subscription and works on the web only; anyone can use existing Tools.
Is Google Flow free?
You can use Flow without a subscription with 50 credits a day, enough for a few short Veo 3.1 Lite or Fast clips, plus image generation, the Flow Agent, and existing Tools. During peak hours, roughly 2 PM to 5 PM UTC, users without a subscription may not be able to generate videos. Subscriptions add monthly credits, 1080p upscaling, and the ability to create Tools.
Can I use Google Flow on my phone?
Yes, with limits. Google offers Flow apps for Android and iOS, but the web version has every feature. Scenebuilder, camera position and motion, and batch editing are desktop-only, and creating Tools is web-only.
Last updated: