How much does Gemini Omni Flash video cost?

Billing is based on total token consumption: input tokens covering text, audio and video cost $1.875 per million, and output tokens cost $21.875 per million. For 720p video that works out to approximately $0.13 per second, audio included. Because the meter is tokens, both resolution and duration move the bill.

Does Gemini Omni Flash support text-to-video?

No. Both endpoints are image-driven: image-to-video animates a single still, and reference-to-video accepts text, images, audio and video together as references. For prompt-only generation use Veo 3.1, Kling or Seedance 2.0. A common workaround is to have the agent generate a still with its image model first, then feed that frame to Omni Flash.

Is Gemini Omni Flash the same as Gemini 3.6 Flash?

No. Gemini 3.6 Flash is a chat model that goes in your instance Model slot and answers messages. Gemini Omni Flash is a video generation model that goes in the Video slot. The two slots are independent, so you can run either alongside the other.

What are the Gemini Omni Flash model IDs?

The fal endpoint paths are google/gemini-omni-flash/image-to-video and google/gemini-omni-flash/reference-to-video. In OpenClaw config they carry a fal/ prefix, for example fal/google/gemini-omni-flash/image-to-video.

← Home

Guide

Gemini Omni Flash Video on OpenClaw: Animate Stills with Sound

Gemini Omni Flash is Google's image-driven video model — it turns a still into coherent motion with audio, grounded in Gemini's physical understanding of how scenes behave. It is one of the cheapest ways to get video with sound into an agent, with one significant catch: there is no text-to-video endpoint.

Not the Gemini Flash You Chat With

The naming needs untangling first. Gemini 3.6 Flash is a chat model that goes in your instance's Model slot and answers messages. Gemini Omni Flash, on this page, is a video generation model that goes in the Video slot. The slots are independent, so you can pair either with anything else.

Two Endpoints, Both Image-Driven

Image to video — animates a single still into video with audio, extending one frame into coherent motion.
Reference to video — accepts text, images, audio and video together as combined references, guiding subject, motion, style and sound at once.

Note what is missing: there is no plain text-to-video endpoint. Your agent needs a starting image or a set of references. If you want a clip from nothing but a written prompt, use Veo 3.1, Kling or Seedance 2.0 instead. A common pattern is to let the agent generate a still with its image model first, then feed that frame to Omni Flash.

Pricing Is Token-Based

Unlike most video models in the picker, Omni Flash bills on total token consumption rather than a flat per-second rate:

Meter	Rate
Input tokens (text, audio, video)	$1.875 per 1M tokens
Output tokens	$21.875 per 1M tokens
Practical rate at 720p	approximately $0.13 per second of video

Where that lands it. At roughly $0.13 per second with audio included, Omni Flash undercuts Veo 3.1 with audio ($0.40/sec) and Seedance 2.0 at 720p ($0.3034/sec), while costing more than Grok Imagine v1 ($0.05/sec). Because the meter is tokens, resolution and duration both move the bill — treat $0.13 as a 720p estimate, not a fixed price.

Setting It Up

Create a fal key at fal.ai/dashboard/keys and save it on your API Keys page.
On your instance card, open the Video dropdown and select the fal.ai provider group.
Pick Gemini Omni Flash, or type “Omni” in the dropdown search box — the fal list refreshes daily from trending endpoints, so the visible set shifts.
Send your agent an image and ask it to animate the shot. The MP4 comes back in the conversation.

Model IDs for Self-Hosted OpenClaw

fal/google/gemini-omni-flash/image-to-video
fal/google/gemini-omni-flash/reference-to-video

What's Next?

Veo 3.1 — Google's flagship video family, with real text-to-video
Google Gemini on OpenClaw — the Gemini chat models
fal.ai on OpenClaw — the provider setup behind this model
Make videos with your bot — the end-to-end brief-to-MP4 flow