Google Makes Gemini Omni 1.2 Flash Generally Available

Google has made Gemini Omni 1.2 Flash generally available on the Gemini API and Google Cloud, with native video clips up to 15 seconds — up from 10 seconds on Omni 1.1 — priced at $0.07 per second for 720p video, 30% less than the previous Flash tier, DeepMind said on 5 November 2026 and Google Cloud repeated.

Creators can set duration in one-second steps from 3 to 15 seconds, draft at 360p or 720p then upscale selected clips to 1080p or 4K, and refine the same clip through multi-turn conversational edits instead of regenerating from scratch. Outputs carry an imperceptible SynthID watermark; Cloud also offers Provisioned Throughput for production capacity.

What changed from Omni 1.1

DeepMind’s product framing is volume creative work: product videos, virtual try-ons and ads that need consistent characters, packaging and brand colours across camera moves. Omni 1.2 Flash is pitched as holding image, short motion (up to three seconds) and audio references more tightly across a full 15-second shot. Cloud’s post adds more natural multi-speaker dialogue and lip-sync, with audio references still gated behind allowlisting.

The cost story is not only the headline rate. Drafting cheaply at lower resolution, fixing one detail in a follow-up turn, then upscaling the keeper is the workflow Google is selling — fewer full regenerations per accepted clip. That matters for UK and EU marketing teams already buying API video by the second: the meter is public ($0.07/s at 720p) even if enterprise discounts are not. DeepMind also lists availability across the Gemini API, Google Vids and Gemini for business, so the same model surfaces in consumer-facing editors and developer endpoints on day one of GA.

Cloud’s customer blurbs (Artlist, FLORA, invideo, Photoroom, QuickFrame) are marketing quotes, not independent benchmarks. They do, however, show where Google wants the SKU to land: ad variants, product spins and CTV-length spots that fit inside a single 15-second generation rather than a stitch of shorter clips.

Video models are converging on edit loops, not one-shot clips

The competitive context on TSN is adjacent generative tooling rather than a single rival launch. OpenAI’s Images 2.6 pushed still-image controls; Gemini’s file-editing chat already treated documents as conversational artefacts; our assistant comparison guide is where readers still sort which stack owns which job. Omni 1.2 Flash extends Gemini’s side of that map into longer native video with an edit loop that looks more like a session than a batch job.

That matters because procurement teams often compare “video model A versus B” on a single demo clip. A 15-second native ceiling with mid-clip conversational edits changes the unit of work: one generation can cover a full social or CTV spot length, and a colourway swap need not burn another full render. Whether that is cheaper in practice still depends on how Google meters follow-up turns — a detail the GA posts leave unstated beyond the $0.07-per-second 720p headline.

Fifteen seconds is a product claim, not a quality certificate

Generally available means the SKU is on the menu with Cloud’s production language (SynthID, Provisioned Throughput, SLAs named by some design-platform customers in Cloud’s post). It is still Google’s own characterisation of the 30% price cut and the 15-second ceiling. SynthID watermarking is a provenance signal, not a guarantee against misuse, and Provisioned Throughput is a capacity product for teams that need reserved performance rather than on-demand bursts.

For UK buyers, the immediate checklist is price maths at 720p, whether multi-turn edits bill as new seconds, and which Workspace or Vids seats actually expose the 3–15s step control. Watch next for audio-reference allowlist widening, UK/EU data-region wording in Cloud docs, and independent evals that compare Omni 1.2 Flash against other API video models at the same duration and resolution. Until those shift, the confirmed Gemini Omni update is GA at 15 seconds and $0.07/s for 720p, with multi-turn editing as the operational change.

Sources

• Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips (Google DeepMind, 5 November 2026)
• Gemini Omni 1.2 Flash generally available (Google Cloud, November 2026)

Related stories

• OpenAI Launches Images 2.6 With Flare and Sunburst
• Gemini Now Lets You Edit Documents, PDFs and Slides Inside the Chat
• Grok, ChatGPT, Claude or Gemini? Pick the AI by the Job