Google launched Gemini Omni 1.2 Flash on November 5, 2026, expanding its video model to clips of up to 15 seconds. The company also cut video output prices and opened the model for general use through its developer services.
Gemini Omni focuses on making and editing videos, rather than general chat or coding. Google says users can revise an existing clip through follow-up instructions and keep subjects more consistent across shots.
Gemini Omni pricing falls as clips get longer
Google’s Gemini API price list puts video output at $12 per million tokens, down from $17.50 for Omni 1.1 Flash. Tokens are the units used for billing. The company describes the reduction as 30%, while its listed input and text-output prices remain unchanged.
| Billing item | Omni 1.1 Flash | Omni 1.2 Flash |
|---|---|---|
| Input: text, images and video | $1.50 / million tokens | $1.50 / million tokens |
| Text output, including thinking | $9 / million tokens | $9 / million tokens |
| Video output | $17.50 / million tokens | $12 / million tokens |
| 720p video, approximate output cost | $0.10 / second | $0.07 / second |
Source: Google Gemini API pricing
Data as of: 2026-11-05
Video output becomes cheaper; listed input and text-output rates stay unchanged.
The same price list estimates output costs of about $0.70 for a 10-second 720p clip and $1.04 for 15 seconds. Those are video-output estimates, not a total bill including input and text output. Google bills actual token use, so multiplying the rounded seven-cent rate can give a slightly different result.
Google lists approximate output rates of $0.023 per second at 360p, $0.104 at upscaled 1080p and $0.209 at upscaled 4K. The Gemini API model has no free tier.
What the video model can generate and edit
Google’s model documentation lists text, images and video as inputs, with video, native audio and text as outputs. Clips run at 24 frames per second. The documentation lists output options from 360p through 4K; its guide identifies 1080p and 4K as upscaled resolutions.
The Gemini Omni model page lists an input limit of 131,072 tokens and an output limit of 65,536 tokens. For readers comparing Google’s models, Likiweeks previously covered Gemini 4 Carbon’s coding and reasoning capabilities. Omni is a separate video release.
Google’s Omni guide sets different limits for different tasks. A clip created from a prompt can be longer than a sequence built between two supplied frames, as the table shows.
| Task | Duration | Important limit |
|---|---|---|
| Generate or edit a clip | 3–15 seconds | Six reference images maximum |
| Connect a first and last frame | 3–10 seconds | The 15-second limit does not apply |
| Extend a clip | 3–10 seconds per turn | 30 seconds total; uses final 10 seconds as context |
| Use reference video for new generation only | Up to 3 seconds per reference | Not for edits or extensions; one reference for 11–15-second output, up to three for 3–10-second output |
Source: Google Gemini Omni guide
Data as of: 2026-11-05
The 15-second maximum applies to generation and editing, with shorter limits for other tasks.
What Google says has improved
According to Google’s launch announcement, follow-up edits can replace a product or background, change lighting or weather, and switch aspect ratios. The company says the model better preserves faces, clothing and packaging details as scenes change. These are Google’s reported improvements, rather than results from an independent comparison.
Google also reports smoother dialogue, better matching of mouths to speech and more control over continuous camera movement. Its examples include product try-ons and scenes with several speakers. The company says fast sports action, sudden script changes and complex physics still need work.
Where Gemini Omni is available
Google’s launch announcement says developers can try the model in Google AI Studio and build with the Gemini API. It also lists Flow and Google Ads among the products getting Gemini Omni. In the Gemini app, 15-second video generation is rolling out to Google AI Ultra subscribers.
The announcement says Google Vids offers 15-second clips at no additional cost with a Google account or Workspace subscription. That app access is separate from the paid developer API.
Google Cloud’s own announcement says the model is generally available on Cloud, with reserved capacity through Provisioned Throughput. However, that post describes access inside Gemini for business as coming soon. Its wording is narrower than the main launch post, which describes availability across Gemini for business.
Gemini Omni audio and regional limits
The Cloud post says audio references need approval through an allowlist. Only approved users can access that feature. The separate Gemini API guide says uploading audio inputs or audio references is unsupported. Generating a clip with sound therefore does not mean any user can upload a voice sample to guide it.
Google’s Gemini API guide also sets limits for users in the European Economic Area, Switzerland and the United Kingdom. They cannot edit or extend uploaded videos. Editing or extending videos generated by the model remains supported. The guide further bars voice editing and some uploads involving minors or recognizable people.
The same guide says Gemini Omni fully supports English. Google has not tested other languages. It also says generation time depends on clip length, resolution and current demand. Google has not supplied a fixed completion time in that guidance.
Watermarking and the switch from Omni 1.1
Google Cloud says outputs carry an invisible SynthID watermark. Google’s API guide says software can detect this marker to check the video’s origin. It also says safety filters apply to prompts and generated clips.
Google’s API changelog says paid-tier rate limits have doubled with the release. It also sets December 3, 2026, as the date when requests to Omni 1.1 Flash will automatically route to Omni 1.2 Flash and use the new model’s rates.
Sources: Google launch announcement, Google API pricing, Google model documentation, Google video guide, Google Cloud, Google API changelog
