Google releases Gemini Omni 1.2 Flash with 15-second clips and a 30% lower price

Google has made Gemini Omni 1.2 Flash generally available, releasing the video model on 5 November 2026 through the Gemini API and Gemini for business. Google Cloud lists it as available the same day, and it also reaches people using Google Vids and Flow. Google points to Flow for scenes with several characters talking, Gemini for business for explainers, and Vids for avatar videos with music and narration. The update centres on longer clips, conversational editing, closer adherence to reference material and a lower price.

Clips can now run up to 15 seconds natively, up from 10 seconds in Omni 1.1, with duration set in one-second steps from 3 to 15. Google prices the model at $0.07 per second at 720p, 30% below the previous version, which puts a full 15-second clip at about $1.05. One report notes that Google's materials give no complete price table by resolution, and a search snippet from another outlet cites a wider $0.02 to $0.10 range per second.

“Beyond the lower price, the workflow keeps costs down: you can draft fast at 360p or 720p, fix a detail with a follow-up edit instead of re-generating the entire clip, and then upscale your best creations to 1080p or 4K.”

Anish Nangia, Product Manager, Google DeepMind, in Google's launch post · blog.google ↗

Inputs can mix character, product or style images, motion clips of up to three seconds and audio cues, though audio references are still limited to an allowlist on Cloud. Google's showcase example turns one model photo plus front and back shots of three outfits into a 15-second virtual try-on with 360-degree turns. Editing is multi-turn: users can swap a product or background, change lighting or weather, or change the aspect ratio without regenerating the clip. Dialogue is described as more natural, with clearer lip-sync and distinct voices for several speakers, and camera moves such as FPV flythroughs and orbits are meant to need fewer jump cuts. A demo flew through the dome of the Pantheon in a single shot.

The Cloud listing covers text-to-video, image-to-video, first and last frame interpolation and extension, with up to three videos per prompt. It supports the Interactions API but not GenerateContent, and only the global region. Outputs carry an imperceptible SynthID watermark and Provisioned Throughput is available. Google pitches the model at product videos, try-ons, catalogue spins, explainers and social ad variants made at scale. QuickFrame AI's chief technology officer, Richard Girges, is reported to see finished 15-second connected-TV ads in a single pass as a significant step for advertising workflows.

Independent testing is thin so far. A Segmind run of 20 API calls saw clips finish in 28 to 136 seconds, mostly between 40 and 90, with the model strongest on explainers, typography and single edits and weaker on cinematic ads and photoreal people. MLQ points out that Google has not published a benchmark or latency comparison against Omni 1.1 for consistency, lip-sync or physics, and Google says work continues on fast sports action, rapid script changes and complex physics.

The release follows the first Gemini Omni Flash, announced at Google I/O on 19 May 2026, and Omni 1.1 Flash on 27 August. Google says image and audio output are planned for the Omni family over time.