Google now generates video through Gemini Omni rather than Veo, and the change affects which plan you need, what Flow credits buy, which API model to call and when older models stop working. Gemini Omni is a Google model family that generates output from any input, starting with video, announced at Google I/O on 20 May 2026 with Omni Flash as its first model. Flow is Google's AI creative studio, introduced at I/O 2025, where Omni and the remaining Veo 3.1 models are used to make and edit clips. This page sets out the Omni versions, how access works in the Gemini app, Flow, YouTube and the Gemini API, the credit and token pricing, the documented limits, the Veo retirement dates and how SynthID and C2PA apply. It is written for creators, marketing teams, developers and the security and governance staff who approve AI tools.
Last checked: 2026-11-15.
This is an independent reference page from NHI Mgmt Group based on public provider documentation; it is not sponsored or endorsed by any provider.
Key takeaways
- Omni Flash 1.2 is the current consumer model. Google's Gemini Apps help names Gemini Omni 1.2 Flash as its latest video model, and Flow lists Omni Flash 1.2 as current.
- The API model is gemini-omni-1.1-flash. The Gemini API model page lists it as stable, with gemini-omni-flash-preview shut down on 22 October 2026.
- Veo 3.1 is being retired in stages. Veo 3.1 preview models left the Gemini API on 22 October 2026, and Flow retires Veo 3.1 Fast, Veo 3.1 Quality and Omni Flash 1.1 on 5 December 2026.
- Flow runs on credits. Free users get 50 credits a day; Google AI Plus, Pro and Ultra add 200, 1,000 and 10,000 or 25,000 credits a month.
- Every Omni output carries SynthID. Google says Omni videos carry an imperceptible SynthID watermark that can be checked in the Gemini app.
What are Gemini Omni and Flow?
Google describes Gemini Omni as a model that combines Gemini's intelligence with its generative media models and can merge image, text, video or audio references into one output. At launch only voice references were supported for audio. Google says it plans to extend Omni to other output types over time. Flow is the studio app; Google says it has grown into an AI creative studio with video and image tools and launched in over 140 countries.
Key facts
| Field | Gemini Omni (as of 2026-11-15) | Google Flow (as of 2026-11-15) |
|---|---|---|
| Provider | ||
| What it is | Video generation and editing model family | AI creative studio app for video and images |
| Announced | 20 May 2026 (Google I/O 2026) | Introduced at I/O 2025 |
| Current version | Omni Flash 1.2 (apps); gemini-omni-1.1-flash (Gemini API) | Uses Omni Flash 1.2, Veo 3.1 Lite and Nano Banana image models |
| Output | Video with generated audio | Video and images |
| Where available | Gemini app, Flow, YouTube Shorts Remix, YouTube Create, Gemini API | Over 140 countries |
| Open weights | No | Not applicable |
| Watermarking | SynthID on all outputs | Inherits model watermarking |
Models and versions
| Version | Where documented | Status (as of 2026-11-15) |
|---|---|---|
| gemini-omni-flash-preview | Gemini API deprecations page: released 30 June 2026 | Shut down 22 October 2026; replacement gemini-omni-1.1-flash |
| Omni Flash 1.1 (gemini-omni-1.1-flash) | Gemini API model page, latest update August 2026; Flow models page | Stable in the Gemini API; retiring in Flow on 5 December 2026 |
| Omni Flash 1.2 | Gemini Apps help and Flow models page | Current in the Gemini app and Flow |
Google's I/O post on 20 May 2026 referred to Omni Flash without a version number. The Gemini API documentation does not list a 1.2 identifier as of 15 November 2026, so developers and app users are on different versions.
Which option fits which need?
These are factual criteria from Google's documentation, not a ranking.
- Occasional clips with no subscription: Flow's free tier gives 50 daily credits and 3 to 10 second Omni Flash 1.2 clips; YouTube Shorts Remix and YouTube Create offer Omni at no cost to adults.
- Longer clips, extensions and 4K: need a Google AI subscription in Flow, with 4K upscaling on Ultra.
- Applications and automation: the Gemini API with gemini-omni-1.1-flash, billed per token with no free tier.
- Enterprise cloud deployment of Veo: Google Cloud lists Veo 3.1 GA models with a retirement date of 4 May 2027 or later.
- Work accounts: the Gemini app requires a qualifying Workspace licence; see the Gemini for Google Workspace Security Guide.
For Omni beside other video models with identical fields, see AI Video Generators Compared.
Pricing and plans
Flow credits
| Plan | Credits |
|---|---|
| No subscription | 50 credits per day; no rollover; video generation may be unavailable at peak hours (about 2 PM to 5 PM UTC) |
| Google AI Plus | 200 extra credits per month |
| Google AI Pro | 1,000 extra credits per month |
| Google AI Ultra $100 | 10,000 extra credits per month |
| Google AI Ultra $200 | 25,000 extra credits per month |
As of 15 November 2026, Google's credits page lists these costs. Omni Flash 1.2 videos up to 10 seconds cost 1 credit per second at 360p and 3 credits per second at 720p. Seconds 11 to 15 cost 2 credits per second at 360p and 6 at 720p, for subscribers only. Upscaling to 720p or 1080p is free for the first 10 seconds, and 4K costs a flat 50 credits for up to 10 seconds on Ultra. Veo 3.1 Lite costs 5 credits per clip for Ultra subscribers and 10 for everyone else. Monthly credits do not roll over. Subscribers can buy top-up AI credits in supported regions except Japan, and those may expire after a period stated at purchase. Google notes that limits can change at any time.
Gemini API pricing
| Item | gemini-omni-1.1-flash (Standard tier, as of 2026-11-15) |
|---|---|
| Free tier | Not available |
| Input (text, image, video or audio) | $1.50 per 1M tokens |
| Text output | $9.00 per 1M tokens |
| Video output | $17.50 per 1M tokens |
| Per second of 720p video | 5,792 tokens, roughly $0.10 per second |
| Batch, Flex or Priority | Not listed for this model |
Features
The API model page lists a 1,048,576 token context window, text, image and video input, video output of 3 to 10 seconds, 360p, 720p, 1080p and 4K resolutions and 24 FPS. Calls go through the Interactions API, returning the video as base64 or, for long videos, as a URI.
| Capability | Documented behaviour |
|---|---|
| Text-to-video | Audio generated by default and steerable by prompt |
| Image-to-video | Reference image; first and last frame interpolation with two images; subject references from several images |
| Editing | Conversational multi-turn editing via previous_interaction_id; uploaded videos via the Files API, 10 seconds or less |
| Extension | Appends to the end of a clip only, up to 40 seconds in total |
| Video references | Up to 3, each 3 seconds or less |
| Resolution | 720p default; 1080p and 4K are upscaled |
| Aspect ratios | 16:9 (default) and 9:16 |
| Not supported | Voice editing, audio reference uploads, multi-video prompting, system instructions, temperature, top_p, stop sequences, negative prompts, YouTube videos as sources, provisioned throughput |
- Regional rules: editing or extending uploaded videos is unavailable in the EEA, Switzerland and the UK; editing model-generated videos is supported there.
- People: uploading and editing images of certain recognisable people is not supported, and images containing minors cannot be uploaded or edited in the EEA, Switzerland or the UK.
- Safety filters: applied to prompts and generated videos, varying by region.
- Language: languages other than English are not evaluated and results may vary.
Availability
- Gemini app: personal accounts need a Google AI plan; work and school accounts need a qualifying Workspace licence. The feature is not available to users under 18.
- Flow: all users can use Omni Flash 1.2, with longer clips and extra features for subscribers (see the table below).
- YouTube: Google said users aged 18 and over can use Omni at no cost in YouTube Shorts Remix and the YouTube Create app.
- Gemini API: gemini-omni-1.1-flash through the Interactions API, paid tier only.
Gemini app limits
Google's help page gives videos of 3 to 10 seconds, or up to 15 seconds for AI Ultra subscribers. One video and up to five images can be uploaded per generation, a generated video can be extended by up to another 10 seconds, and multi-turn editing keeps revising a video in the same conversation. Users can record their face and voice to create an avatar and call it with @ and their Google username. Video-to-video editing and avatar uploads are not available in the EEA, Switzerland, the UK or some US states. Videos may be removed if Google's systems detect a possible breach of its Terms of Service or Prohibited Use Policy.
Flow features by plan (Omni Flash 1.2)
| Capability | No subscription | Google AI Plus, Pro or Ultra |
|---|---|---|
| Text, frames, ingredients and video-to-video | 3 to 10 seconds, both aspect ratios | 3 to 15 seconds, both aspect ratios |
| Omni Extend | Not included in the free tier | 3 to 10 second extensions, up to 3 images, web only |
| Upscaling Omni 360p video | Not included in the free tier | To 1080p; to 4K on Ultra |
| Omni 360p | One third of the 720p cost | One third of the 720p cost |
Flow also offers Flow Agent, which Google says handles multi-step tasks rather than one prompt at a time and is available to all Flow users, and Flow Tools, custom creative tools built in natural language that subscribers can create and remix.
Provenance and labelling
- SynthID: Google's API guide says all Omni outputs include SynthID watermarking, invisible to viewers but detectable programmatically.
- Verification: SynthID verification for image, video and audio is in the Gemini app, where Google says it had been used 50 million times by I/O. Google said it was expanding to Search at launch and to Chrome in the following weeks, through Lens, AI Mode, Circle to Search and Gemini in Chrome.
- C2PA: Content Credentials verification rolled out in the Gemini app from 20 May 2026, with Search and Chrome to follow. The Gemini API Omni pages do not mention C2PA, while the Google Cloud Veo 3.1 page lists C2PA support.
- Partners: Google said OpenAI, Kakao and ElevenLabs are bringing SynthID to more of their AI-generated content.
A watermark shows where a file came from; it does not prove a clip is harmless or that a person consented to appear in it. For how the two schemes differ, see SynthID vs C2PA and the AI Content Provenance and C2PA Guide.
Data and privacy
Google's Gemini API pricing page lists gemini-omni-1.1-flash as available on the paid tier only, and marks paid-tier content as not used to improve Google's products. Use in the Gemini app, Flow and YouTube falls under Google's consumer terms and activity settings rather than the API terms, so check those settings separately for staff who use the consumer apps.
Where identity and security fit
Omni's avatar feature binds a person's recorded face and voice to their Google account, which makes consent and account security part of content risk. Gemini API keys used for Omni are machine credentials that should be scoped, rotated and removed when a project ends. Staff using personal Google AI plans for work video create shadow AI exposure; the AI Image and Video Generation at Work guide covers the policy side, and the Deepfake and AI Impersonation Guide covers misuse of generated likenesses.
Release history
Google announced Gemini Omni at Google I/O on 20 May 2026. The API preview, gemini-omni-flash-preview, was released on 30 June 2026 and shut down on 22 October 2026 in favour of gemini-omni-1.1-flash. Omni Flash 1.2 is the current version in the Gemini app and Flow. Flow itself was introduced at I/O 2025.
When does Veo 3.1 retire?
Veo retirement differs by surface. The Gemini API deprecations page says its shutdown dates are the earliest possible dates and exact dates are communicated in advance.
| Surface and model | Released | Retirement | Replacement or note |
|---|---|---|---|
| Gemini API: veo-3.1-generate-preview, veo-3.1-fast-generate-preview | 15 October 2025 | 22 October 2026 (past) | gemini-omni-1.1-flash |
| Gemini API: veo-3.1-lite-generate-preview | 31 March 2026 | 22 October 2026 (past) | gemini-omni-1.1-flash |
| Gemini API: veo-3.0-generate-001, veo-3.0-fast-generate-001, veo-2.0-generate-001 | 9 September 2025 (Veo 3.0); 9 April 2025 (Veo 2.0) | 30 June 2026 (past) | gemini-omni-1.1-flash or GA models on Gemini Enterprise Agent Platform |
| Flow: Veo Extend | Not stated | 5 November 2026 (past) | Extending needs Omni Flash 1.2 and a paid plan |
| Flow: Veo 3.1 Fast, Veo 3.1 Quality, Omni Flash 1.1 | Not stated | 5 December 2026 | Omni Flash 1.2 or Veo 3.1 Lite |
| Google Cloud: veo-3.1-generate-001, veo-3.1-fast-generate-001 (GA) | 17 November 2025 | 4 May 2027 or later | us-central1 |
| Google Cloud: veo-3.1-lite-generate-001 (Preview) | 2 April 2026 | Not listed | us-central1 |
On Google Cloud, the Veo 3.1 GA models generate clips of 4, 6 or 8 seconds at 24 FPS in 16:9 or 9:16, up to four videos per prompt, with English prompts, and support C2PA Content Credentials. Teams that need Veo beyond the Gemini API and Flow dates should check the Google Cloud model page before planning around it. Retirement dates across providers are tracked in the AI Media Model Retirements Tracker.
How do you move off Veo 3.1?
- Find usage: search code for veo-3.1-*-preview, veo-3.0-* and veo-2.0-* identifiers, and ask Flow users which projects rely on Veo 3.1 Fast or Quality.
- Map features: Veo clips were 4, 6 or 8 seconds; Omni Flash 1.2 in Flow runs 3 to 15 seconds and Omni in the API 3 to 10 seconds, with different reference and extension rules.
- Re-cost: Flow moves to per-second credits for Omni Flash 1.2, and the API moves to token billing at about $0.10 per second of 720p video.
- Pilot: regenerate representative shots, check regional restrictions for EEA, Swiss and UK users, and confirm outputs still carry the provenance your publishing rules need.
- Roll back if needed: on Google Cloud, Veo 3.1 GA models remain listed until at least 4 May 2027; in Flow, Veo 3.1 Lite stays available.
Limits
- Clip length: API videos run 3 to 10 seconds, with extension up to 40 seconds in total; Gemini app videos run up to 10 seconds, or 15 seconds on AI Ultra.
- Resolution: 720p is the API default, and 1080p and 4K are upscaled.
- Regional rules: editing uploaded videos and avatar uploads are unavailable in the EEA, Switzerland and the UK.
- Version split: the API offers gemini-omni-1.1-flash while the apps run Omni Flash 1.2.
- Languages: Google says languages other than English are not evaluated and results may vary.
Practitioner checklist
- Inventory Flow projects and API code that use Veo 3.1 Fast, Veo 3.1 Quality or Omni Flash 1.1 before 5 December 2026.
- Replace any remaining veo-3.1 preview or gemini-omni-flash-preview calls with gemini-omni-1.1-flash.
- Decide which staff need a Google AI plan or Workspace licence, rather than relying on personal accounts.
- Budget Flow credits per second of output and track top-up purchases, which may expire.
- Check EEA, Swiss and UK restrictions on uploaded-video editing and avatars before rolling out to those teams.
- Set a policy on avatars and recognisable people, including consent records.
- Keep SynthID and C2PA signals intact in publishing pipelines and add visible AI labels where rules require them.
- Scope and rotate Gemini API keys used for video generation.
Standards and references
- Google: I/O 2026 announcements
- Google AI for Developers: Gemini Omni Flash model page
- Google AI for Developers: Generate and edit videos with Gemini Omni Flash
- Google AI for Developers: Gemini API deprecations
- Google AI Studio: Pricing
- Google Gemini Help: Generate videos with Gemini Apps
- Google Flow Help: Learn about Google Flow models and supported features
- Google Flow Help: Manage your Google Flow credits
- Google Cloud: Veo 3.1 model page
Related NHI Mgmt Group resources: AI Image, Video, Voice and Music Models Hub · Google Gemini Models · Google Nano Banana Explained · OpenAI GPT Image Models Explained