Gemini Omni Flash
Live now on RunwareGoogle's multimodal video model for generation and conversational editing
Access Gemini Omni Flash, Google's latest video generation model.
All workflows supported natively.
One integration for generation and editing, with no separate endpoints.
Gemini Omni Flash pricing
Metered per token on input and text output, and per second on video output. No subscription, no minimums — same key and billing as every other model on Runware.
Input tokenizes at 2,040 tokens per image, 32 tokens per second of audio, and 5,792 tokens per second of video. Video output is billed per second of 720p output with audio rather than per token.
Why Gemini Omni Flash stands out
Grounded in real-world knowledge
An intuitive grasp of physics, plus Gemini's knowledge of history and science as well as culture, for believable scenes.
Location & world awareness
Understands real places and generates regionally accurate settings, down to local signage and language.
Character consistency across scenes
One reference image keeps a character or product consistent across scenes, angles, and lighting.
Character & object swaps
Replace characters or objects just by asking. Motion and dialogue carry over seamlessly.
Motion & style transfer
Apply motion from a source clip to a reference character, or restyle footage without regenerating the scene.
Synced text and audio
On-screen text and sound that connect coherently to what's happening in the frame.
Built for real workflows
Access Gemini Omni Flash, Google's multimodal model that creates and edits video from text, image, video, and audio inputs, grounded in Gemini's real-world knowledge. Iterate through natural conversation, with every edit building on the last while keeping your scene consistent. It's built for real production work, not just isolated clips.
Storyboard to finished video
Upload a storyboard image and get a full video with dialogue and narrative.
Character-led storytelling
Turn a single character image into a multi-scene story with narration and captions.
Background & setting swaps
Drop a subject into a new environment, like a Bollywood dance scene.
Location-aware generation
Generate regionally correct outputs from real places, coordinates, signage, and native language.
Natural-language editing
Change one segment with a follow-up instruction, like swapping Sydney for the Taj Mahal.
Production at scale
Turn a few assets into many production-grade outputs fast.
How it compares
Where Gemini Omni Flash sits next to other leading video models. Its edge is conversational, multi-turn editing on top of generation, grounded in Gemini's real-world knowledge.
| Model | Gemini Omni Flash | Veo 3.1 | Kling VIDEO 3.0 Omni Standard | Seedance 2.0 | HappyHorse 1.0 |
|---|---|---|---|---|---|
| Max duration | 10s | 8s | 15s | 15s | 15s |
| Input types | Text, image, video | Text, image, video | Text, image, video | Text, image, video, audio | Text, image, video |
| Reference inputs | Up to 7 images | Up to 3 images | 7 images + 1 video + 7 elements | 9 images + 3 video + 3 audio | Up to 9 images |
| Native audio | Yes | Yes | Yes | Yes | No |
| Max resolution | 720p | 4K | 720p | 4K | 1080p |
| Capabilities | Text-to-video, image-to-video, video-to-video, edit | Text-to-video, image-to-video, audio-to-video, edit, extend | Text-to-video, image-to-video, edit | Text-to-video, image-to-video, video-to-video, audio-to-video, edit, extend | Text-to-video, image-to-video, edit |
Get the most out of Gemini Omni Flash
Guides covering cinematic prompting, reference-driven video, and editing, written for building with Gemini Omni Flash on Runware.
Cinematic prompting
How to prompt Gemini Omni Flash for cinematic video using Google's five-element structure, camera language, and the less-prescriptive sweet spot.
Reference-driven video
How to use the reference image workflow to lock a visual style, hold a character across scenes, or guide a video through storyboard key beats.
Editing video
How to edit existing footage with the inputs.video parameter to relight, restyle, swap weather, or add characters while preserving the source's composition and motion.
Get started
One API call, in whichever language you're already using — same key and billing as every other model on Runware.
curl https://api.runware.ai/v1 \
-H "Authorization: Bearer $RUNWARE_API_KEY" \
-H "Content-Type: application/json" \
-d '[
{
"taskType": "videoInference",
"taskUUID": "<uuid-v4>",
"model": "google:gemini@omni-flash",
"positivePrompt": "A slow push-in on a violinist practicing alone in an empty concert hall, dust motes drifting in a shaft of afternoon light. Sound: solo violin, room reverb, a page turning.",
"duration": 5
}
]'[
{
"taskType": "videoInference",
"taskUUID": "<uuid-v4>",
"status": "success",
"videoUUID": "1e4f7a92-6c3b-4d81-9a05-8b2e6f1c7d39",
"videoURL": "https://vm.runware.ai/video/ws/2/vi/1e4f7a92-6c3b-4d81-9a05-8b2e6f1c7d39.mp4",
"cost": 0.5
}
]Frequently asked questions
What is Gemini Omni Flash?
Google's multimodal model that generates and edits video from text, image, video, and audio inputs, grounded in Gemini's real-world knowledge. In Google's own shorthand: “Think of Gemini Omni Flash like Nano Banana, but for video.” It's a conversational, iterative editing workflow, not just a generation tool.
What makes Gemini Omni Flash different?
Conversational, multi-turn editing where each change builds on the last while keeping the scene consistent, plus an intuitive grasp of real-world physics, drawing on Gemini's knowledge of history and science as well as culture.
What workflows does it support?
Text-to-video, image-to-video, reference-to-video, and multi-turn video editing, all through one Runware integration.
How does Gemini Omni Flash compare to other models?
Most video models focus on single-shot generation. Gemini Omni Flash adds conversational, iterative editing on top of generation, so you can refine a scene over multiple turns instead of regenerating from scratch, while keeping characters and continuity consistent. On Runware it sits alongside other leading video models, so you can pick the right one for each task through the same API.
How much does Gemini Omni Flash cost?
Input tokens (text, image, video, and audio) are billed at $1.50 per 1M tokens. Text output costs $9.00 per 1M tokens. Video output is $0.10 per second of 720p video. On the input side, Google charges 2040 tokens per image and 5792 per video second, with audio at 32 tokens per second. Full pricing is listed on the Runware pricing page.
Do I need separate integrations?
No. Generation and editing run through the same API.
Is Gemini Omni Flash available now?
Yes. Gemini Omni Flash is live on Runware. You can try it directly in the Playground using your existing Runware credentials and billing.
Already integrated with Gemini Omni Flash somewhere else?
Our Migration Skills and docs help map an existing integration over — same request shape as every other Runware model, so the change is usually just field names and auth. See the migration guide for details.
Talk to us about volume discounts
Have questions about Gemini Omni Flash? Chat to our team about enterprise usage, including volume discounts and dedicated RPM, and we'll follow up shortly.