Veo 3
Veo3 is Google DeepMind's video generation model, the first major model to produce dialogue, sound effects and music natively.
Try Veo 3 online → Full guide on GitHub
Overview
Veo 3 is Google DeepMind's video generation model, released in May 2025 as the first major model to generate audio natively, including lip-synced dialogue, sound effects and music. It produces eight-second 720p or 1080p clips from text or an image, with Veo 3 Fast as a cheaper variant and Veo 3.1 adding reference images, first-and-last-frame control, scene extension and 4K. It is available in the Gemini app, Flow, YouTube Shorts, the Gemini API and Vertex AI.
What it can do
- Native audio with lip-synced dialogue
- 8 second clips at 720p, 1080p or 4K (3.1)
- Text-to-video and image-to-video
- Up to 3 reference images for consistency (3.1)
- First-and-last-frame transitions
- Scene extension for longer sequences
Versions
| Version | Released | Notes |
|---|---|---|
| Veo 3 | 2025-05 | Announced at Google I/O; native audio, 8 second 720p clips, 16:9. Launched in the Gemini app for Ultra subscribers, Flow and Vertex AI preview. |
| Veo 3 Fast and image-to-video | 2025-07 | Cheaper, faster variant; image-to-video in the Gemini app and API; Veo 3 arrives in the paid Gemini API preview. |
| Veo 3 GA, 1080p and vertical | 2025-09 | Stable API model IDs veo-3.0-generate-001 and veo-3.0-fast-generate-001; 1080p output and 9:16 vertical; price cuts. |
| Veo 3.1 and Veo 3.1 Fast | 2025-10 | Reference images, first-and-last-frame with audio, scene extension, richer audio and better prompt adherence; Flow gains insert and remove. |
| Veo 3.1 update | 2025-12 | 4K output, vertical video in the API and longer extensions. |
How to try Veo 3
Veo 3 is available in the Gemini app and Flow for Google AI Pro and Ultra subscribers, in YouTube Shorts as Veo 3 Fast, and to developers through the Gemini API in Google AI Studio and Vertex AI, billed per second of video with no free tier. Country availability, daily quotas and the newest 3.1 features depend on the plan. The quickest route without a subscription or waitlist is the pay-per-generation service at https://veovideoai.com.
FAQ
What is Veo 3?
Veo 3 is Google DeepMind's video generation model, released in May 2025 as the first major model to generate synchronized audio, including dialogue, sound effects and music, alongside eight-second video clips from text or images.
Is Veo 3 free?
Mostly not. YouTube Shorts offers a free, lower-resolution Veo 3 Fast, but the Gemini app and Flow require Google AI Pro or Ultra, and the Gemini API has no free tier for Veo.
Is there a Veo 3 API?
Yes. Veo 3, Veo 3 Fast and Veo 3.1 are available in the Gemini API through Google AI Studio and on Vertex AI, billed per second of output, and are also resold by hosts such as fal.ai and Replicate.
Does Veo 3 have an official GitHub repository?
No. Veo 3 is a closed, hosted model and Google has not released weights or an official repository. This page collects publicly available information about it.
How do I try Veo 3 online?
Subscribe to Google AI Pro or Ultra and use the Gemini app or Flow, or use the Gemini API. For pay-per-generation access without a subscription or waitlist, use https://veovideoai.com.
What are the limits of Veo 3?
Each generation is at most eight seconds, so longer videos rely on extension and stitching; 1080p and 4K are tied to certain aspect ratios; safety filters block real public figures and some depictions of people; and quotas and availability vary by country and plan.
What is the difference between Veo 3 and Veo 3.1?
Veo 3.1 keeps the eight-second base clip and native audio but adds up to three reference images for consistency, first-and-last-frame transitions with sound, scene extension, richer audio, better prompt adherence and, in a later update, 4K output.