If you generate video with Google's models, October 22, 2026 matters. Google's Gemini API deprecations page lists the Veo 3.1 preview models (veo-3.1-generate-preview, veo-3.1-fast-generate-preview and veo-3.1-lite-generate-preview) with a shutdown date of October 22, 2026. The recommended replacement on the Gemini API is Gemini Omni Flash (gemini-omni-1.1-flash), released August 27, 2026. Generally available Veo models continue on Google's enterprise platform.
What is Gemini Omni Flash?
Google describes Omni Flash as a multimodal model "designed for high-speed video generation, editing, and cinematic control." It takes text, images and video as input, generates video with an audio track by default, and supports conversational editing: each request can build on the previous result.
Gemini Omni Flash vs Veo 3.1
Generation modes
Both handle text-to-video, image-to-video, first and last frame interpolation and reference images. Omni adds editing of generated or uploaded videos and multi-turn refinement.
Resolution and format
- Veo 3.1: 720p, 1080p or 4K (4K not on Lite); 16:9 or 9:16; 4, 6 or 8-second clips; 24 fps.
- Gemini Omni Flash: 360p, 720p (default), with 1080p and 4K listed as upscaled; 16:9 or 9:16.
Audio
Veo 3.1 audio is always on. Omni generates an appropriate audio track by default; Google notes that uploading audio references and voice editing are not supported in the current version.
References
Veo 3.1 accepts up to three reference images of a person, character or product. Omni supports subject references from multiple images and up to three short video references (up to 3 seconds each).
Extension
Veo 3.1 (Standard and Fast) extends a clip by 7 seconds at a time, up to 20 times. Omni generates continuations of a few seconds at a time; Google's documentation gives slightly different figures in different sections, so check current limits before planning long clips.
Prompting differences
- Omni creates several shots by default. For one shot, write "In a single continuous shot, no scene cuts."
- For edits, keep prompts short and add "Keep everything else the same."
- Omni understands timecodes such as [0-3s] and tags like
<FIRST_FRAME>. - Veo responds well to quoted dialogue and explicit sound descriptions. See our Veo 3.1 prompt guide.
What should creators do?
- If you use Veo through a platform, the platform decides which Google endpoint it calls. Existing prompts should keep working, but results may change if a platform switches models.
- Keep your best prompts and re-test them on Omni and on Veo's enterprise version where available.
- Use Omni for iteration (edit and refine) and Veo for final high-resolution output where both are offered.
Want to compare models side by side? Try the same prompt across the video models in the TextVidAI generator.
FAQ
Is Veo 3.1 being shut down?
The Veo 3.1 preview model IDs on the Gemini API have a shutdown date of October 22, 2026. Generally available Veo models remain on Google's enterprise platform.
What replaces Veo 3.1 on the Gemini API?
Google lists gemini-omni-1.1-flash as the recommended replacement.
Does Gemini Omni generate sound?
Yes, it generates an audio track by default.
Sources
- Google: Gemini API deprecations
- Google: Gemini Omni Flash documentation
- Google: Veo 3.1 documentation
Last updated: October 9, 2026.