Gemini Omni Flash 1.1
Google's video model, metered by the second
- referenceImages
- frameImages
- referenceVideos
- video
Gemini Omni Flash is Google's flash-tier video generation model - the fast, cheap end of the same family whose premium end is Veo-class - and this node serves it over Runware's cloud with a native VIDEO output. "Omni" is the whole story: it generates synchronized audio along with the picture rather than bolting sound on afterward, which has been the defining upgrade of the 2026 video generation. And it's the only way you'll run it, because Google's models have never shipped weights. This is a closed model you rent, and the node is deliberately shaped to make that painless and legible.
Legible is doing a lot of work there. The defaults tell you the intended use: size starts at 360p ~16:9 - not a typo, and not a quality insult so much as a cost discipline. Flash-tier video is for iterating on ideas fast and cheap, and the node's defaults are pointed at that job. duration defaults to 6 seconds. If the output feels small, that's the trade you signed up for by picking the Flash tier; bump the size preset and watch the per-run price climb on the node's title bar.
The inputs are the standard current-gen set with one addition that marks this as a Google model. Alongside referenceImages, frameImages, referenceVideos, and a video input for editing or extending an existing clip, you get settings.temperature (default 1) and settings.topP (default 0.95) - the two LLM-style sampling knobs that don't usually appear on diffusion video nodes. Lower temperature if you want more deterministic, on-brief results; raise it when the output is coming out safe and you want variety. positivePrompt is required and sentence-style. numberResults, the safety group with its fast/full moderation modes, outputFormat (MP4/WEBM/MOV) and outputQuality complete the list. Output: VIDEO.
Install
The one Runware pack setup covers every node here: ComfyUI Manager → search Runware → install → restart, or git clone https://github.com/Runware/ComfyUI-Runware into ComfyUI/custom_nodes plus pip install -r requirements.txt. Then a key from the Runware dashboard - Settings → Runware API key, RUNWARE_API_KEY, or runware auth login.
The honest framing
This node exists for the "I can't run it locally" category, and Google video is about as can't-run-locally as it gets. The things to keep an eye on: per-call cost on video is real and adds up, your prompt and reference clips leave the machine into Google's moderation, and the model's own content policy is welded in at the source - no setting here will un-refuse what Google refuses. Use the 360p default for drafts, treat the sampling knobs as your main creative control, and save the big sizes for the take you're actually keeping.
Inputs (17)
| Name | Type | Default | Description |
|---|---|---|---|
| positivePrompt | STRING | Text prompt describing elements to include in the generated output. | |
| referenceImagesopt | IMAGE | — | |
| frameImagesopt | IMAGE | — | |
| referenceVideosopt | RUNWARE_REFERENCEVIDEOS | — | |
| videoopt | STRING | Video input (UUID or URL). Source video to edit or extend. | |
| sizeopt | COMBO | 360p (16:9) | Output size. Only one of resolution or dimensions is sent. |
| durationopt | INT | 63–10 | Length of the generated video in seconds. The total number of frames produced is determined by duration multiplied by the model's frame rate (fps). |
| numberResultsopt | INT | 11–4 | Number of results to generate. Each result uses a different seed, producing variations of the same parameters. |
| safetyopt | BOOLEAN | false | Enable to set safety. Off uses the model's default. |
| safety.checkContentopt | BOOLEAN | false | Enable or disable content safety checking. Increases total generation time. |
| safety.modeopt | COMBO | fast | Safety checking mode for video generation. |
| settings.temperatureopt | FLOAT | 1.000–2 | Controls randomness in generation. Lower values produce more deterministic outputs, higher values increase variation and creativity. |
| settings.topPopt | FLOAT | 0.950–1 | Nucleus sampling parameter that controls diversity by limiting the probability mass. Lower values make outputs more focused, higher values increase diversity. |
| ttlopt | BOOLEAN | false | Enable to set ttl. Off uses the model's default. |
| ttl_valueopt | INT | 60 | Time-to-live (TTL) in seconds for generated content. Only applies when `outputType` is `URL`. |
| outputFormatopt | COMBO | MP4 | File format for the generated video. |
| outputQualityopt | INT | 9520–99 | Compression quality of the output. Higher values preserve quality but increase file size. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| video | VIDEO | — |