Google DeepMind · Video · Updated October 1, 2026Gemini API preview endpoints shut down 22 Oct 2026

Veo 3.1: the consistency model

Google DeepMind's Veo 3.1 is the video model to reach for when the same product or presenter has to look identical across five cuts. Here is what it does, what it costs per second, and where the SynthID watermark and 8-second base clip get in a marketer's way.

What Veo 3.1 is

Veo 3.1 is Google DeepMind's flagship video generation model, released in mid-October 2025, producing text-to-video and image-to-video clips with synchronized 48kHz audio at native 1080p (upscaled to 4K). Its defining feature is Ingredients to Video: up to three reference images keep a character, product or object consistent across scenes. Base clips are about 8 seconds, extendable to roughly 60, and every output carries Google's SynthID watermark.

For short-form marketers it is the most reliable way to keep a product or spokesperson looking the same across a batch of Reels without a reshoot, at $0.05 to $0.60 per second depending on tier. Note that Google is shutting down the Veo 3.1 preview endpoints in the Gemini API on 22 October 2026 and now points developers to Gemini Omni.

Veo 3.1 specs

Maker	Google DeepMind
Released	Mid-October 2025 (exact day not verified in available sources)
Max clip length	About 8 seconds per generation, extendable to roughly 60 seconds of continuous footage
Resolution	Native 1080p, AI upscaling to 4K
Native audio	Yes: 48kHz synchronized dialogue, effects and ambience
Modes	Text-to-video, image-to-video
Consistency	Ingredients to Video: up to 3 reference images for a consistent character, product or object across scenes and angles
Frame rates	24, 30 or 60fps
Pricing	Gemini API, with audio: Lite $0.05 per second at 720p and $0.08 at 1080p (no 4K); Fast $0.10 at 720p, $0.12 at 1080p and $0.30 at 4K; Standard $0.40 at 720p or 1080p and $0.60 at 4K. Vertex AI bills per generation
Status	Google's Gemini API deprecations page lists veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview for shutdown on 22 October 2026, with gemini-omni-1.1-flash as the recommended replacement. The Vertex AI model (veo-3.1-generate-001) is unaffected by that date; its retirement timing is not verified
Watermark	SynthID, visible and embedded, on every output

What changed in Veo 3.1

Versus Veo 3.0, 3.1 added Ingredients to Video for multi-reference consistency, improved frame-to-frame consistency (reported 40 to 60% better in internal testing), added 4K upscaling, and refined audio quality and sync. Veo 3.1 is still the current Veo as of 1 October 2026: Google has not shipped or announced a Veo 4. Google's newer video work at I/O 2026 went out under the Gemini Omni name rather than as a Veo release, and the Gemini API video docs now recommend Gemini Omni Flash as the default, keeping Veo 3.1 for legacy use until its preview endpoints shut down on 22 October 2026.

How to get Veo 3.1

Available in the consumer Gemini app, Google Flow, the Gemini API for developers, Vertex AI for enterprise, and third-party platforms such as Higgsfield and Freepik. Pricing differs meaningfully across those paths. The Gemini API preview endpoints shut down on 22 October 2026; Vertex AI and the apps are not affected by that date.

What Veo 3.1 means for short-form marketers

Ingredients to Video is the feature that matters. A product shot, a mascot or a presenter can be held consistent across a batch of cuts, which is the thing most video models fail at and the thing a feed full of variants needs.

The trade-offs are the watermark and the base length. SynthID on every frame is a real consideration for UGC-style content that is supposed to look filmed, and 8-second base clips mean the extend workflow for anything longer.

Cost is workable at the Standard tier for testing, but true 4K with audio is the expensive path, and the pricing maze across Gemini, Flow and Vertex makes budgeting harder than it should be.

Where Veo 3.1 still needs care

All outputs watermarked with SynthID, visible and embedded
Base clip is about 8 seconds; longer needs the extend workflow
4K is an upscale, not native, and the most expensive tier
Access and pricing vary across Gemini, Flow, Vertex and third parties
The Gemini API preview endpoints shut down on 22 October 2026, so developer integrations need to move to Vertex AI or to Gemini Omni

Using Veo 3.1 with Fastlane

Generate anywhere, publish natively to TikTok, Reels, Shorts and LinkedIn, attribute signups and sales to every post.

Use Veo 3.1 for the shots that need a consistent product or presenter, then bring the clips into Fastlane to schedule weeks ahead and publish natively to TikTok, Instagram Reels, YouTube Shorts and LinkedIn, with signups and sales attributed to each post.

For the volume around those hero shots, Fastlane generates from your website URL: AI UGC videos with 1,000+ characters, slideshows, hook + demo videos and memes, remixed from 25,000+ trending videos. Free plan, no credit card.

Try Fastlane free See the AI TikTok video generator

Veo 3.1: frequently asked questions

What is Veo 3.1?

Google DeepMind's flagship video generation model, released mid-October 2025, with native audio, 1080p output and a multi-reference consistency feature called Ingredients to Video.

How much does Veo 3.1 cost?

Through the Gemini API, with audio: Lite costs $0.05 per second at 720p and $0.08 at 1080p, Fast $0.10 to $0.30 per second, and Standard $0.40 per second at 720p or 1080p and $0.60 at 4K. Vertex AI bills per generation.

Does Veo 3.1 have a watermark?

Yes. Every output carries Google's SynthID watermark, both visible and embedded.

How long can a Veo 3.1 video be?

About 8 seconds per generation, extendable to roughly 60 seconds of continuous footage.

Is Veo 3.1 good for TikTok ads?

It is one of the best options for keeping a product or presenter consistent across variants. Plan around the watermark and the short base clip.

Has Veo 4 been released?

No. As of 1 October 2026 Google has not released or announced a Veo 4. The video model Google launched at I/O 2026 shipped as Gemini Omni Flash, not as a Veo version, and Google now recommends it over the Veo 3.1 preview endpoints, which shut down in the Gemini API on 22 October 2026.

Is Veo 3.1 being shut down?

Partly. The Veo 3.1 preview endpoints in the Gemini API (standard, Fast and Lite) shut down on 22 October 2026, and Google recommends gemini-omni-1.1-flash instead. Veo 3.1 stays available on Vertex AI and in Google's apps for now.

Sources

Specs verified against the sources above on October 1, 2026. Model access and pricing change often; check the maker's page before you build on any number here.

Ship 30 days of viral content in 30 seconds

Join 50,000+ founders and marketers publishing the same video to TikTok, Instagram, YouTube and LinkedIn. Free to start.

Try Fastlane free See pricing