No. As of October 25, 2026, Google has not released Veo 4, and it has not announced one either. Google DeepMind's Veo page still calls Veo 3.1 "our leading video generation model." The Gemini API's Veo page lists nothing newer than Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite, and those preview models were shut down in the Gemini API on October 22, 2026. What Google shipped instead is a different model family: Gemini Omni Flash, which Google's developer docs now recommend as the default for video generation.
So if you searched "Veo 4 release date," the answer is that there isn't one. Below is what exists, what each model actually does according to Google's own documentation, which signals would tell you a Veo 4 is coming, and what to use in the meantime.
An earlier version of this post, published May 20, 2026, described Veo 4 as released. That was wrong. It listed specs, an Oakgen credit price, and side-by-side test results for a model that did not exist. We have removed all of it and rewritten the post from Google's primary documentation. We are sorry for the error.
Google shut down the Veo 3.1 Preview models in the Gemini API on October 22, 2026. GA Veo 3.1 is still on Vertex AI. On Oakgen, HappyHorse (retired September 27, 2026) and Runway Gen-4 Turbo are no longer available, so the Oakgen table below now lists only models Oakgen runs.
Veo 4 status at a glance
Every row below comes from Google's own pages, checked on September 26, 2026. The Gemini API status was rechecked on October 25, 2026. Dates are Google's announcement or release dates.
| Model | Status (Oct 2026) | First released | Clip length | Resolution | Audio | Where Google offers it |
|---|---|---|---|---|---|---|
| Veo 4 | Not announced, not released | n/a | n/a | n/a | n/a | Nowhere |
| Veo 3 / Veo 3 Fast | Deprecated in the Gemini API | May 20, 2025 (Google I/O) | 8 s | 720p, 1080p (16:9 only) | Native, always on | Still listed on Vertex AI |
| Veo 3.1 / Veo 3.1 Fast | Preview shut down in Gemini API Oct 22, 2026; GA on Vertex AI | Oct 15, 2025 | 4, 6 or 8 s (8 s for 1080p, 4K, references) | 720p, 1080p, 4K | Native, always on | Gemini API, Vertex AI, Google Flow |
| Veo 3.1 Lite | Preview shut down in Gemini API Oct 22, 2026 | Mar 31, 2026 | 4, 6 or 8 s | 720p, 1080p (no 4K) | Native, always on | Gemini API, Vertex AI |
| Gemini Omni Flash | GA (gemini-omni-1.1-flash) | Preview Jun 30, 2026; GA Aug 27, 2026 | 3 to 10 s at preview launch | 360p, 720p native; 1080p and 4K upscaled | Native | Gemini API, Vertex AI, Gemini app, Google Vids |
Two details worth knowing. Veo 3.1 and Veo 3.1 Fast went GA on Vertex AI on November 17, 2025, with a listed retirement date of "November 17, 2026 or later." And Google announced on June 15, 2026 that the Veo 3 model IDs in the Gemini API would shut down on June 30, 2026.
What Veo 3.1 actually does
Veo 3.1 is the newest model that carries the Veo name. Google's Gemini API documentation describes it as a model "for generating 8-second videos (720p, 1080p, or 4k) with natively generated audio." Here is the verified spec sheet:
- Durations: 4, 6 or 8 seconds. 1080p, 4K and reference-image jobs must be 8 seconds.
- Resolution: 720p (default), 1080p, and 4K. Google added 4K output on January 13, 2026. Veo 3.1 Lite tops out at 1080p.
- Frame rate: 24fps.
- Aspect ratio: 16:9 or 9:16.
- Audio: always on. Dialogue, sound effects and ambience come from the prompt.
- Reference images: up to three, to guide subject and style.
- First and last frame: you can supply both and let the model fill the motion between them.
- Extension: extend a Veo-made clip, at 720p only, to a combined length of up to 148 seconds.
- Watermark: every output carries a SynthID watermark.
- Retention: generated videos stay on Google's servers for 2 days.
Google's launch post for Veo 3.1 (October 15, 2025) framed it as bringing "richer audio, more narrative control, and enhanced realism," with stronger prompt adherence than Veo 3. We covered the release in our Veo 3.1 4K and native audio breakdown, and compared it with Kling and Seedance in Seedance 2.5 vs Kling 3 vs Veo 3.1.
The limit that trips people up is length. A single Veo 3.1 generation is 8 seconds at most. Anything longer needs extension, and extension drops you to 720p.
What Gemini Omni Flash is (and why it matters for Veo 4)
Gemini Omni Flash is the model Google now points developers to first. The Gemini API's video overview says it plainly: "Use Gemini Omni Flash as your default model for video generation." It keeps Veo 3.1 for "specific capabilities like scene extension, last-frame control, or integration with legacy pipelines."
What Google documents about Omni Flash:
- Model ID: gemini-omni-1.1-flash (GA). The preview endpoint, gemini-omni-flash-preview, is scheduled for deprecation on September 30, 2026.
- Inputs: text, images, audio and video in the same request.
- Conversational editing: through the Interactions API you can ask for changes in plain language across several turns, and the model edits the clip instead of starting over.
- Length: the June 30 preview launch described 3 to 10 second videos. The Gemini app lists "Create 10 second videos."
- Resolution: 720p by default, 360p available, and 1080p or 4K produced by upscaling.
- Extension and interpolation: GA added end-of-clip extension and first-plus-last-frame generation.
- Video references: up to 3 clips, up to 3 seconds each.
- Watermark: SynthID on every output.
The consumer side moved too. The Gemini app's video page says "Gemini Omni will replace Veo in the Gemini app," and describes Omni 1.1 Flash as the model "that will now replace the previous Google Gemini Veo 3.1 model." On September 23, 2026, Google announced that anyone with a Google account can make 1080p videos with Gemini Omni in Google Vids at no cost.
This is the most useful context for anyone waiting on Veo 4. Google's newest video model did not arrive as a Veo version number. It arrived as part of Gemini. That does not rule out a Veo 4, but it does mean the next big jump in Google video may not use that name.
Veo release history (so you can read the calendar yourself)
| Date | Release | Where it was announced |
|---|---|---|
| May 14, 2024 | Veo | Google I/O 2024 |
| Dec 16, 2024 | Veo 2 | Google Labs update |
| May 20, 2025 | Veo 3 (first Veo with native audio) | Google I/O 2025 |
| Oct 15, 2025 | Veo 3.1 and Veo 3.1 Fast | Google blog and Gemini API changelog |
| Mar 31, 2026 | Veo 3.1 Lite (preview) | Gemini API changelog |
| Jun 30, 2026 | Gemini Omni Flash (preview) | Gemini API changelog |
| Aug 27, 2026 | Gemini Omni Flash GA | Gemini API changelog |
Two of the three major Veo versions debuted at Google I/O. Point releases showed up whenever they were ready.
Signals that a Veo 4 announcement is close
None of these is a prediction. They are the places a real launch would show up first, so you can check them instead of trusting a rumor post (including, until today, this one).
- A new model ID in the Gemini API docs. Veo 3.1, Veo 3.1 Lite and Gemini Omni Flash each showed up in the Gemini API changelog with a model ID, and they are listed on the Gemini API models page. A Veo 4 without a model ID is marketing, not a release.
- The Google DeepMind Veo page headline changes. Today it says Veo 3.1 is the leading video generation model. A new version would have to replace that line.
- Deprecation notices for Veo 3.1. After Veo 3.1 arrived, Google set shutdown dates for the older Veo 3 IDs. Vertex AI lists Veo 3.1's retirement as "November 17, 2026 or later," so watch that page.
- Google I/O. Veo and Veo 3 both launched on I/O keynote day. If Google keeps that pattern, the next I/O keynote is the obvious date to watch.
- Google Flow and Gemini app release notes. Veo 3 launched alongside Flow, and Veo 3.1 launched with a batch of Flow updates. Since the Gemini app has already moved to Omni, a new Veo appearing there would be a strong sign.
- Naming. If the next announcement is "Gemini Omni" something rather than "Veo 4," Google has answered the question for you.
If a site claims Veo 4 specs, look for a link to a Google domain (deepmind.google, ai.google.dev, cloud.google.com or blog.google) that says the same thing. If there isn't one, the specs were made up.
What to use today instead of Veo on Oakgen
Oakgen does not offer Veo anymore. We removed our Veo models in late August 2026, and old Veo model links now redirect to the AI video generator. If you came here to generate a "Veo 4" clip on Oakgen, these are the models that cover the same jobs.
Credit costs below come from Oakgen's pricing table, as of October 25, 2026. The generator shows the credit price before you run.
| Job you had in mind for Veo | Model on Oakgen | Clip length on Oakgen | Resolution | Native audio option | 5-second clip |
|---|---|---|---|---|---|
| Google's own newer video model | Gemini Omni Flash | Short clips | 720p | Not listed | 169 credits |
| Longer single takes, reference-heavy scenes | Seedance 2.5 | 4 to 30 s | 480p, 720p, 1080p | Yes | 373 credits (480p), 800 (720p), 1,968 (1080p) |
| Drafts and 4K finals | Seedance 2.0 | 4 to 15 s | 480p up to 4K | Not listed | 228 credits (480p), 512 (720p) |
| Premium realism, dialogue with sound | Kling v3 Pro | 3 to 15 s | 1080p | Yes | 284 credits with audio, 190 without |
| Cheap drafts, references, sound | MiniMax H3 Max | 5 to 15 s | 480P, 768P, 1080P | Yes | 85 credits (480P), 136 (768P), 271 (1080P) |
A few notes on picking between them:
- If you liked Veo because it is Google's, Gemini Omni Flash is Google's newer model and runs on Oakgen at 720p.
- If you liked Veo for sound, start with Kling v3 Pro, Seedance 2.5 or MiniMax H3 Max, which all have an audio option. Our Seedance 2.5 guide covers what its references and audio can and can't do.
- If you liked Veo for 8-second cinematic shots, Seedance 2.5's 4 to 30 second range removes the need to stitch clips, which was Veo 3.1's biggest constraint.
- If you are animating a product photo, MiniMax H3 Max at 480P is the cheapest test at 85 credits for 5 seconds, and Grok Imagine Video costs the same at 480p. Our walkthrough on how to animate a photo compares approaches.
For a wider view, see our Veo 3 alternatives roundup and Seedance 2.5 vs Kling 3 vs Veo 3.1. Treat their Veo sections as a comparison point, not an Oakgen option.
Make the Clip You Wanted From Veo 4
Run the same prompt through Gemini Omni Flash, Seedance 2.5 and Kling v3 Pro from one credit balance.
A quick way to test a Veo replacement
Don't pick a model from a table. Pick it from your own prompt. This takes about ten minutes:
- Write one prompt for the shot you actually need, including subject, action, camera move, lighting and any line of dialogue.
- Run it on two models at the lowest resolution each offers. On Oakgen, MiniMax H3 Max at 480P (85 credits) and Seedance 2.0 at 480p (228 credits) keep a 5-second test near 300 credits.
- Check three things only: does the subject stay the same person, does the camera do what you asked, and is the audio usable (if you need it).
- Re-run the winner at your delivery resolution. Only then move to a premium model like Kling v3 Pro if the cheaper one missed.
Testing at full resolution is the fastest way to burn credits. Draft cheap, finish expensive.
Sources
All accessed September 26, 2026. The Google DeepMind Veo page and the Gemini API Veo page were rechecked on October 25, 2026.
- Google DeepMind, Veo model page: deepmind.google/models/veo
- Google DeepMind, Gemini Omni model page: deepmind.google/models/gemini-omni
- Gemini API, Video generation overview: ai.google.dev/gemini-api/docs/video
- Gemini API, Generate videos with Veo 3.1: ai.google.dev/gemini-api/docs/veo
- Gemini API, Generate and edit videos with Gemini Omni Flash: ai.google.dev/gemini-api/docs/omni
- Gemini API, Models list: ai.google.dev/gemini-api/docs/models
- Gemini API, Release notes: ai.google.dev/gemini-api/docs/changelog
- Google Cloud, Veo 3.1 model page (Vertex AI): cloud.google.com/vertex-ai/generative-ai/docs/models/veo/3-1-generate
- Google Cloud, Model list: cloud.google.com/vertex-ai/generative-ai/docs/models
- Gemini app, Gemini Omni video generation page: gemini.google/overview/video-generation
- Google blog, Introducing Veo and Imagen 3 (May 14, 2024): blog.google
- Google blog, Veo 2 announcement (December 16, 2024): blog.google
- Google blog, Veo 3, Imagen 4 and Flow (May 20, 2025): blog.google
- Google blog, Introducing Veo 3.1 (October 15, 2025): blog.google
- Google blog, Gemini Omni in Google Vids (September 23, 2026): blog.google
- Oakgen model configuration and pricing table for Gemini Omni Flash, Seedance 2.5, Seedance 2.0, Kling v3 Pro and MiniMax H3 Max, as of October 25, 2026.
Frequently asked questions
Is Veo 4 out?
No. As of October 25, 2026, Google has not announced or released a model called Veo 4. Google DeepMind's Veo page still describes Veo 3.1 as its leading video generation model, and the Gemini API's Veo page lists nothing newer than Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite.
When will Veo 4 come out?
Google has not given a date, and nobody outside Google can give you a real one. Past Veo launches landed at Google I/O (Veo in May 2024, Veo 3 in May 2025), in December (Veo 2) and in October (Veo 3.1). Google's newest video model, Gemini Omni Flash, shipped under the Gemini name rather than as a new Veo version, so a future upgrade may not be called Veo 4 at all.
Is Veo 3.1 the latest Veo model?
Yes. Veo 3.1 and Veo 3.1 Fast were released in October 2025, and Veo 3.1 Lite followed in preview on March 31, 2026. Veo 3 and Veo 3 Fast are marked deprecated in the Gemini API. Google's developer docs now recommend Gemini Omni Flash as the default video model and keep Veo 3.1 for features like scene extension and last-frame control.
What is Gemini Omni Flash, and is it Veo 4?
Gemini Omni Flash is Google's multimodal video generation and editing model. It entered public preview in the Gemini API on June 30, 2026 and became generally available as gemini-omni-1.1-flash on August 27, 2026. Google does not call it Veo 4. The Gemini app says Omni will replace Veo in the app.
Can I use Veo on Oakgen?
No. Oakgen removed its Veo models in late August 2026, and old Veo model pages now redirect to the main AI video generator. For similar jobs, Oakgen offers Gemini Omni Flash (Google's newer video model), Seedance 2.5, Seedance 2.0, Kling v3 Pro and MiniMax H3 Max.
What is the best alternative to Veo for video with sound?
Among the models Oakgen offers, Kling v3 Pro, Seedance 2.5 and MiniMax H3 Max have a native audio option. Seedance 2.5 supports clips from 4 to 30 seconds at 480p, 720p, or 1080p, and Kling v3 Pro supports 3 to 15 seconds. Run the same prompt through both before committing a project to either.
Where can I use Veo 3.1 today?
Not in the Gemini API any more. Google's Veo page says the Veo 3.1 Preview models in the Gemini API were shut down on October 22, 2026, and points developers to Gemini Omni Flash or the GA Veo models. GA Veo 3.1 and Veo 3.1 Fast remain on Google Cloud's Vertex AI. Check Google's current pages for pricing and plan requirements.


