AI · · 5 min read

Gemini Omni 1.2 Flash Goes GA: 15-Second Clips, Conversational Edits, 30% Cheaper

Google DeepMind's production video model stretches native generation to 15 seconds, lets you refine the same clip turn by turn, and drops 720p pricing to $0.07 per second.

Soft editorial illustration of three muted video frames linked by a wavy ink ribbon toward a terracotta orb on warm paper
Illustration by TechPulse Editorial · One clip, many turns, fewer regenerations.

Google DeepMind put Gemini Omni 1.2 Flash into general availability on November 5, 2026, and the pitch is blunt: longer clips, edits that do not throw away the take, and a lower price for the volume work businesses actually ship.

What changed from Omni 1.1

Length. You can generate and edit videos natively up to 15 seconds, up from 10 seconds on Omni 1.1. Duration can be set in one-second increments from 3 to 15, which is enough for a product explainer or a full social ad spot in one generation.

Editing. Multi-turn conversational editing keeps the same clip in play. Google’s examples swap backgrounds, products, lighting, weather, or aspect ratio without re-briefing the whole shot. That matters for ad variants: one hero generation, many placements.

References. The model accepts character, product, or style images, short motion clips (up to three seconds), and audio cues, and Google says it holds faces, garments, packaging typography, and brand colors more tightly across camera angles. Audio references remain allowlist-gated.

Speech and camera. Launch posts call out more natural multi-speaker dialogue with clearer lip-sync, plus steadier continuous moves (FPV flythroughs, orbits, crash zoom-outs) with fewer unwanted cuts.

Cost. 720p generation is $0.07 per second, described as about 30% cheaper than Omni 1.1 Flash, with similar savings at other resolutions. The recommended workflow is draft at 360p or 720p, iterate with conversational edits, then upscale keepers to 1080p or 4K.

Where you can call it

Omni 1.2 Flash is live for developers on the Gemini API (model id gemini-omni-1.2-flash in Google’s cookbook), for businesses in Gemini for business and the Google Cloud console with Provisioned Throughput, and inside Google apps including Flow, Google Vids, the Gemini app (rolling out to AI Ultra), and Google Ads. Outputs carry an imperceptible SynthID watermark.

Google is frank about remaining weak spots: fast sports action, rapid script changes, and complex physics are still improving. For catalog spins, virtual try-on, explainers, and branded short-form, the GA release is the production path.

Why developers should care

Video models usually punish iteration: every tweak burns another full generation. Conversational editing plus a draft-then-upscale ladder changes the unit economics for agentic creative pipelines that generate dozens of variants per SKU. If you already orchestrate image and copy agents, Omni 1.2 Flash is the piece that can stay on the same thread instead of restarting the brief each time.

Sources & Primary Advisories

  1. Google — Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips
  2. Google Cloud — Gemini Omni 1.2 Flash now GA
  3. Google DeepMind — Gemini Omni 1.2 Flash model card
Share Dispatch