Gemini Omni Flash 1.1
Google DeepMind's conversational video model — generate from text, images, or references, edit existing footage in plain language, and deliver up to 4K.
curl https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-omni-flash-1-1-text-to-video",
"prompt": "Aerial drone shot over a misty mountain valley at golden hour",
"duration": "4s",
"resolution": "720p",
"aspect_ratio": "16:9"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/video/retrieve.
# Call /video/complete after downloading if needed.Overview
What is Gemini Omni Flash 1.1
Gemini Omni Flash 1.1 is Google DeepMind's conversational video generation and editing model, released August 27, 2026. It creates clips from text, still images, reference images, or existing footage, with resolutions up to 4K and 4–10 second durations. On Venice it runs under an anonymized privacy tier — prompts are never stored.
Using it anonymously on Venice
On Venice, Gemini Omni Flash 1.1 runs under the anonymized privacy tier: your prompts are not stored, profiled, or tied to a personal generation history — a sharp contrast to Big-Tech platforms that retain every prompt by default. You pay per clip instead of subscribing, so a 360p draft costs $0.16 and you only scale up to 4K when a shot earns it. All four modes — text-to-video, image-to-video, reference-to-video, and video-to-video — are available permissionlessly through the app or the Venice API.
Specifications
Datasheet
- Maker
- Google DeepMind
- Modality
- Text-to-video, image-to-video, reference-to-video, video-to-video
- License
- Proprietary
- Clip lengths
- 4s – 10s in 2s steps
- Resolutions
- 360p, 720p, 1080p, 4k
- Mode
- text-to-video
- Audio
- No
- References
- Reportedly up to 14 images and up to 3 reference videos for guiding generation
- Reference video
- Up to 10 seconds per reference clip, up to 3 clips
- Released
- August 27, 2026 (Gemini Omni family debuted at Google I/O, May 2026)
- Architecture
- Transformer-based with native multimodal support for text, image, video, and audio inputs
- Aspect ratios
- 16:9, 9:16
- Privacy on Venice
- Anonymized — prompts not stored
- Available on Venice since
- Aug 2026
Assessment
Strengths and limitations
- Four input modes in one family: prompt a clip from text, animate a still image, bind reference images to characters or products, or restyle an existing video.
- Conversational editing: the video-to-video variant lets you describe changes (swap a character, relight a scene, alter the angle) instead of re-generating from scratch.
- Full resolution ladder from cheap 360p drafts to 4K delivery, with per-clip pricing that makes iteration inexpensive.
- First and last frame interpolation: attach a starting and ending frame and the model generates the footage in between.
- Strong world understanding and physics reasoning inherited from the Gemini backbone, per Google's model card.
- No native audio on Venice: clips are silent, unlike Grok Imagine 1.5 or Vidu Q3 which generate synchronized sound.
- Clips are capped at 10 seconds per generation on Venice; the 40-second scene extension available in Google's own API is not offered here.
- Closed and proprietary: no open weights, so self-hosting or fine-tuning is impossible. Wan 2.7 is the open-weights alternative on Venice.
- Only two aspect ratios: 16:9 and 9:16.
- Third-party reviewers note prompt-only cinematic camera shots are less consistent than dedicated cinematic models.
Use cases
What it is good for
- 01Iterative ad creative: draft at 360p for cents, refine conversationally, deliver the winner at 4K.
- 02Product shots where a reference image keeps the product identical across scenes.
- 03Animating still photography or artwork into motion for social feeds in 9:16.
- 04Restyling or editing existing footage by describing the change in plain language.
- 05Keyframed shots where you control both the first and last frame.
Prompting
Getting better results
Describe camera movement, lighting, and mood explicitly — Google's own docs recommend this for best text-to-video results.
Iterate at 360p · 4s ($0.16) and only re-render the winning prompt at 1080p or 4K.
For reference-to-video, attach reference images and assign roles so characters or products carry over without the image being used as a frame.
For first/last frame work, supply two ordered images — a clean start frame and a distinct end frame.
Pick 9:16 for Reels/Shorts/TikTok and 16:9 for cinematic or YouTube formats before generating, since only these two ratios exist.
For video-to-video edits, phrase the change as an instruction ('swap the jacket to red, keep everything else') rather than re-describing the whole scene.
Samples
Sample outputs
Generated on Venice with our standard prompt suite — the same prompts we run through every model of this type, so you can judge it like-for-like.
Slow aerial drone shot gliding over a misty mountain valley at golden hour, sunlight piercing clouds onto a winding river, ancient pine forests on either side, ultra-smooth motion, professional color grading, atmospheric haze, 4K
Calm ocean waves rolling onto a black-sand beach at sunrise, golden light on wet sand, foam dissolving into the shore, a single silhouetted figure at the waterline, smooth continuous forward push, serene cinematic atmosphere
A slow tracking shot through a rain-soaked Tokyo street at night, neon reflecting in puddles, steam rising from a food cart, a person with a translucent umbrella, shallow depth of field, teal-and-orange grade, smooth steady camera
Alternatives
How it compares
| Model | Best for | Max duration | Max resolution | Native audio |
|---|---|---|---|---|
| Gemini Omni Flash 1.1 | Iterative editing & 4K delivery | 10s | 4K | No |
| Wan 2.7 Enhanced | Open-weights workflows | 15s | 1080p | Yes |
| Grok Imagine 1.5 | Audio-driven social clips | 15s | 1080p | Yes |
| Vidu Q3 | Longest single-shot clips | 16s | 1080p | Yes |
Gemini Omni Flash 1.1 is the right pick when the workflow is iterative — draft cheap at 360p, edit clips conversationally, and deliver at 4K. If a single shot needs native audio or more than 10 seconds, Grok Imagine 1.5 or Vidu Q3 serve better.
API
Call it from your code
Venice exposes this model through the REST API. Queue a generation with the model id.
curl https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gemini-omni-flash-1-1-text-to-video",
"prompt": "Aerial drone shot over a misty mountain valley at golden hour",
"duration": "4s",
"resolution": "720p",
"aspect_ratio": "16:9"
}'
# Use the returned queue_id with https://api.venice.ai/api/v1/video/retrieve.
# Call /video/complete after downloading if needed.Pricing
What it costs on Venice
Pay per clip on Venice — price scales with resolution and duration (4s–10s), from $0.16.
New Venice accounts include a free daily allowance and 500 welcome credits — no credit card required.
FAQ
Frequently asked questions
Gemini Omni Flash 1.1 is Google DeepMind's conversational video generation and editing model, released August 27, 2026 as part of the Gemini Omni family shown at Google I/O 2026. It generates clips from text, images, or reference assets, edits existing footage through natural language, and outputs resolutions up to 4K.
Venice bills per clip, with price scaling by resolution and duration. A 360p 4-second clip costs $0.16, 720p 4 seconds is $0.47, and 1080p 4 seconds is $0.70, with longer 6–10 second clips priced proportionally. There is no subscription required.
It is neither free-software nor open-source: the weights are proprietary to Google DeepMind and cannot be self-hosted. On Venice you can try it with free welcome credits, then pay per clip. If you need open weights for video, Wan 2.7 and Wan 2.7 Enhanced are the open alternatives on Venice.
Not on Venice. While Google's own API version of the model produces a synchronized audio track, the Venice deployment outputs silent video only. If you need native audio, Grok Imagine 1.5, Vidu Q3, or Wan 2.7 Enhanced all generate sound on Venice.
Yes — the gemini-omni-flash-1-1-image-to-video variant animates a still image into motion, with the same 4–10 second clip lengths and 360p-to-4K resolutions as the text-to-video mode. A related image-to-video task also supports first and last frame interpolation, where the model generates the footage between two keyframes.
Yes. The gemini-omni-flash-1-1-reference-to-video variant binds reference images to roles so characters, products, and objects from your images appear in new scenes without the image itself being used as a frame. This is the mode to use for recurring brand characters or product consistency.
Gemini Omni Flash 1.1 wins on editing and delivery: it is the only one of the two that restyles existing footage conversationally and outputs 4K. Grok Imagine 1.5 wins on clip length (15s vs 10s) and native audio. Choose Omni for iterative, edited, high-resolution work; choose Grok for longer sound-on social clips.
Venice runs it under the anonymized privacy tier: prompts are not stored, profiled, or used for training, and generations are not tied to a personal account history. You can use it in the Venice app or through the Venice API without building a data trail.
Four tiers: 360p for cheap drafts, 720p as the everyday workhorse, 1080p for production, and 4K for final delivery. Clip lengths run 4 to 10 seconds in 2-second steps, in 16:9 or 9:16 aspect ratios.
Use Gemini Omni Flash 1.1 anonymously
Venice does not store your prompts. Chat history stays in your browser.