Google Flow is the AI filmmaking tool from Google Labs, built on top of the Veo video models. Where the Gemini app gives you one-off clips, Flow is aimed at people assembling scenes: it keeps characters and locations consistent across shots, lets you extend clips into sequences, and exposes camera controls that plain prompting does not. You will find it at labs.google/fx/tools/flow — searches for "flow ai", "google labs flow", and "labs.google flow" all lead to the same product.
Flow matters because it is the primary consumer doorway to Veo. When people ask how to use Veo 3 or Veo 3.1 "in Flow", they are really asking how Google packages its video model for creators — and what it costs, since Flow sits behind Google’s AI subscription plans.
This guide covers what Flow does, how it uses Veo, how to get access, and where it falls short — including when running Veo models directly on Veo AI, with pay-as-you-go credits instead of a subscription, is the more practical route.
What is Google Flow?
Flow is an AI video creation tool that Google introduced at I/O in May 2025 and has been expanding since. It was designed with filmmakers in mind: instead of generating isolated clips, you work on a project timeline — Flow calls them scenes — where each shot can reuse the same characters, props, and style. Under the hood, the video itself is generated by Google’s Veo models, with Imagen handling still "ingredients" and Gemini interpreting your prompts.
Flow is part of Google Labs, which means it is an experiment-tier product: capabilities, pricing, and limits change frequently, and features arrive in the US first before rolling out to other countries.
How Flow uses Veo 3 and Veo 3.1
Every video Flow produces comes from a Veo model — Flow is the interface, Veo is the engine. When Veo 3.1 shipped, Flow adopted it for its highest-quality generations, including native audio: dialogue, ambience, and effects generated in the same pass as the picture.
That is why searches like "veo 3 flow" and "google flow veo 3" are really about one workflow: writing a prompt in Flow and choosing the Veo quality tier. The same models — and newer ones — are also available outside Flow. On Veo AI you can run Veo 3.1, Veo 3.1 Fast, and Veo 4 directly, alongside Sora, Kling, and 40+ other models, without a Google subscription.
- Text to video: describe a shot and Flow renders it with a Veo model, sound included.
- Ingredients to video: upload reference images — a character, a prop, a location — and Flow keeps them consistent across shots.
- Frames to video: give a first and last frame, and Veo interpolates the motion between them.
- Scene builder: extend a clip or jump to a new camera angle while preserving continuity.
- Camera controls: specify moves like push-ins, pans, or aerial perspectives per shot.
How to access Google Flow
Flow lives at labs.google/fx/tools/flow — typing "google flow" into search and clicking the Labs result gets you there, and the misspelled variants people search for (labs google flow, google.flow, lab.google flow) all resolve to that address. You sign in with a regular Google account.
Full access requires a paid plan: Flow is included with Google AI Pro and, with the highest limits and earliest model access, Google AI Ultra. Free accounts in supported countries can usually try a limited number of generations, but sustained use — and Veo’s best quality tiers — sit behind the subscription. Availability also varies by country, with new features landing in the US first.
Is Google Flow free?
Not really. Flow runs on a monthly AI-credit allowance tied to your Google plan: generations consume credits, higher-quality Veo renders consume more, and when the allowance runs out you wait for the next cycle or upgrade. The free tier, where available, is a trial-sized allowance rather than a working quota.
If what you actually want is Veo output without a monthly commitment, that is the gap Veo AI fills: new accounts get free credits to start, and after that you buy credits only when you need them — no subscription required to keep access to the models.
Google Flow vs Veo AI
Both products generate video with Veo models; the difference is packaging. Flow is Google’s own scene-building environment, deeply integrated with its ecosystem and subscriptions. Veo AI is a multi-model studio: the Veo family next to Sora 2, Kling, Hailuo, Wan, and dozens more, with per-generation credit pricing.
| Google Flow | Veo AI | |
|---|---|---|
| Video models | Veo (current Google release) | Veo 4, Veo 3.1 + 40 other video & image models |
| Pricing | Google AI Pro / Ultra subscription | Free starter credits, then pay-as-you-go credits or plans |
| Scene continuity tools | Yes — scene builder, ingredients | Reference images & keyframes per generation |
| Model choice | Veo only | Compare Veo against Sora, Kling, Hailuo side by side |
| Watermark | SynthID (invisible) on outputs | Clean downloads at full quality |
| Access | Google account + plan, country-dependent | Browser sign-up, works wherever you are |
Which should you use?
Choose Flow if you live inside Google’s ecosystem, already pay for AI Pro or Ultra, and want its scene-builder workflow for longer narrative pieces. It is genuinely good at multi-shot continuity.
Choose Veo AI if you want Veo’s output without the subscription, need to compare models to pick the right engine per job, or want clean full-quality downloads with predictable per-clip costs. Prompts transfer between the two with no changes — a prompt written for Flow runs on Veo 3.1 here as-is.