Google DeepMind Releases Gemini Omni 1.2 Flash With 15-Second Video

Google DeepMind on Nov. 5, 2026, made Gemini Omni 1.2 Flash generally available — a video generation model that produces and edits clips with synchronized audio, now up to 15 seconds long. The release comes with a 30% price reduction from the previous version and a set of workflow changes designed to make iterative video production faster and less expensive at scale.

What’s Actually New

The headline upgrade is clip length. Omni 1.1 topped out at 10 seconds; Omni 1.2 Flash now reaches 15 seconds, selectable in one-second increments starting at three seconds. That range matters because a 15-second clip maps directly to a standard social ad slot — something no previous Flash-tier model offered at a published, production-ready price point.

The API price lands at $0.07 per second for 720p video. That works out to roughly $1.05 for a full 15-second clip. Developers can draft at 360p or 720p, iterate using conversational editing, and then upscale finished clips to 1080p or 4K — keeping costs down during the revision phase.

Multi-turn conversational editing is the other substantive change. Rather than submitting a new prompt every time a detail needs adjusting — swapping a background, changing lighting or shifting aspect ratio — users can refine the same clip through follow-up instructions in a single session. That’s a workflow shift, not just a capability bump: it moves AI video editing closer to how people already work in tools like Figma or Google Docs, where revision happens in context rather than from a blank slate.

Reference adherence has also improved. The model can now hold faces, brand colors, packaging typography and garment details consistent across camera angles and motion transitions when given character images, short motion clips of up to three seconds, or audio cues as reference inputs. Google also says natural-sounding dialogue, clearer lip-sync and multi-speaker conditioning are stronger in this version, alongside more controllable camera movements with fewer unwanted jump cuts.

Google acknowledges that fast sports action, rapid script changes and complex physics remain areas under development — worth noting if those are your specific use cases.

Where Gemini Omni Stands Competitively

The AI video market has changed sharply in 2026. OpenAI’s Sora — once the most-talked-about video model — shut down its app in April 2026 and pulled its API on Sept. 24, 2026. That leaves Runway Gen-4.5 and Kling 3.0 as the most direct live comparisons. Runway Gen-4.5 caps clips at 10 seconds and prices at approximately $0.12 per second. Gemini Omni 1.2 Flash now undercuts that on both length and cost.

Gemini Omni’s core competitive argument has never been raw photorealism — Sora and Runway historically led on cinematic fidelity. The advantage here is workflow integration: video generation and editing within the same API ecosystem where developers are already building with Gemini for text and image tasks, at a cost that makes prototyping practical rather than prohibitive. Kling 3.0 competes on price per generation but lacks Gemini Omni’s native multimodal input pipeline.

Omni 1.2 Flash also becomes the first model in its family to reach general availability, backed by Google Cloud enterprise infrastructure, SLA-grade uptime guarantees, and Provisioned Throughput for high-volume deployments. All outputs carry a SynthID watermark — an imperceptible provenance signal embedded at generation time.

What This Means for Students

Access is tiered in a way that creates a genuine on-ramp for students at every budget level. Anyone with a standard Google account or a Google Workspace subscription — which covers most students already using Gmail or Google Docs — can generate up to three 15-second clips per month in Google Vids at no additional cost. For a class presentation, portfolio piece or social content experiment, that’s a usable free allocation without touching a credit card.

For students building apps or experimenting with AI-video pipelines, a 15-second clip via the API now costs about $1.05 at $0.07 per second. When Sora Pro was running $3 to $5 per clip, that kind of experimentation wasn’t realistic on a student budget. At current Gemini Omni pricing, it is. Google AI Studio also offers a no-cost prototyping environment before committing to API spend.

For anyone eyeing careers in AI product management, marketing technology or media production, the multi-turn editing workflow and reference adherence features are increasingly appearing in job descriptions for AI content roles. Social-volume use cases — YouTube Shorts, branded ad variants, product explainers — are exactly where hiring managers in those fields focus. The SynthID watermarking is also worth understanding directly: it’s a live example of AI provenance tooling that shows up in responsible AI and media ethics coursework and, increasingly, in real editorial and legal conversations.

The practical takeaway: test Omni 1.2 Flash this week in Google Vids for free, understand what it can and can’t do, and treat the $1.05-per-clip API price as a low-stakes entry point to building something real.

Source: Google DeepMind