Gemini Omni 1.2 Flash is Google’s video-model update of November 5, 2026: native clips up to 15 seconds, multi-turn editing, and — per the Google DeepMind blog — 30% cheaper than Omni 1.1 at $0.07 per second for 720p. It is generally available on the Gemini API and Google Cloud; outputs carry an imperceptible SynthID watermark.
Versus Omni 1.1, max length rises from 10 to 15 seconds; duration can be set in one-second steps from 3 to 15. Reference images, short motion clips (up to three seconds), and audio cues are meant to hold character, product, and style across camera angles. Google’s workflow tip: draft at 360p/720p, fix details with follow-up edits instead of full regeneration, then upscale winners to 1080p or 4K.
Gemini Omni 1.2 Flash: Length, Price, Consistency
The 30% price cut versus 1.1 and the $0.07/s (720p) entry point are hard figures from the announcement — not third-party estimates. Provisioned Throughput is cited for enterprise. Google positions the model for social ads, product explainers, and virtual try-on; we stick to specified capabilities and do not adopt campaign promises.
Consistency: less drift for characters, objects, and scenes across motion and camera moves. Multi-turn editing allows background, lighting, or aspect-ratio changes on the same clip. Speech: more fluid dialogue, clearer lip-sync, multi-speaker conditioning. Camera: longer FPV/orbit moves with fewer unwanted jump cuts — per the blog, without independent bench numbers beyond the launch page.
Access and Limits
Developers: Google AI Studio, Gemini API, and docs. Business: Google Cloud console. Apps: Flow for scenes and multi-turn; Google Vids with a Google account/Workspace at no extra cost for 15-second clips; the Gemini app for Google AI Ultra subscribers (rolling out); Google Ads for campaign scale. Google notes ongoing work on fast sports action, rapid script changes, and complex physics — known weak spots, not a universal guarantee.
Framing without hype: Gemini Omni 1.2 Flash is a production video API/Cloud update with clear specs (15s, −30% price, multi-turn, SynthID), not a new foundation model and not a market-share claim. The Google blog of November 5, 2026 remains the primary source.
What Teams Should Check Now
Creative and growth teams: reprice budgets at $0.07/s against prior 1.1 spend; plan variants via multi-turn instead of full regenerations. Product/brand: prepare reference sets (character, product front/back, short motion) for try-on and catalog spins. Platform engineering: align Gemini API and Cloud quotas, SynthID requirements, and internal asset policies. Legal/compliance: read watermarking and Gemini API terms — the launch page does not replace contract review.
Related reading on code404zero: OpenAI Images 2.6, OpenAI Pierre, and GPT-6 Astra. Together they show how model releases and API productization stay coupled — here with a video focus rather than text/code.
Technical Snapshot
Native clip length up to 15 seconds (was 10); duration 3–15s in 1s steps. Price: −30% vs Omni 1.1, $0.07 per second at 720p. References: character/product/style images, motion up to 3s, audio cues. Editing: multi-turn on the same clip. Output: SynthID; GA on Gemini API and Google Cloud; Provisioned Throughput available. The launch narrative does not embed a full resolution price table — developer docs and Cloud pricing remain the purchase sources.
Bottom line: On November 5, 2026, Google shipped Gemini Omni 1.2 Flash with 15-second clips, multi-turn editing, and a 30% lower price versus 1.1. For facts, use the DeepMind/Google blog; no invented benchmarks, no hype claims.
Versus the Prior Version Without Marketing Tone
Omni 1.1 topped out at 10 seconds; 1.2 Flash extends that to 15 and allows finer durations. The launch page stresses reference adherence and multi-turn as cost controls, because variants need not be fully regenerated each time. SynthID remains part of the GA statement. Teams with existing 1.1 pipelines should retarget prompt templates, asset buckets, and cost alerts to the new limits and the $0.07/s rate — and read developer docs for resolution pricing beyond 720p.
Procurement takeaway: specs and price are documented; quality comparisons stay internal. code404zero sticks to the Google primary of November 5, 2026 and adds no invented win rates.
Product Surfaces and Rollout
Google lists several entry points in parallel: Gemini API and AI Studio for prototypes, Google Cloud for enterprise including Provisioned Throughput, Flow for scene-level multi-turn work, Google Vids for 15-second clips at no extra cost with a Google account or Workspace, the Gemini app for Ultra subscribers in rollout, and Google Ads for campaign variants. That is a distribution mesh, not a single consumer feature. Teams should clarify which surface is bound by contract and budget — API cents versus Workspace-included usage are different cost models.
Reference adherence remains the technical lever for brand work: character, product, and style references plus short motion and audio cues. Launch examples (virtual try-on, poster build, FPV architecture) illustrate the spec; they do not replace internal QA gates. SynthID is a required note for transparency processes; internal policies on AI-generated video should cover the watermark and retention of prompt/reference history.
Cost Math Without Speculation
$0.07 per second at 720p and −30% versus Omni 1.1 are the announcement figures. Minute-volume planners can rough it out: 15 seconds × $0.07 = $1.05 per clip at 720p — excluding upscale and storage. Multi-turn lowers effective cost only when edits replace full regenerations; otherwise turns add up. Budgets should attach alerts to seconds and turns, not only clip counts.
Limits per Google: fast sports action, rapid script changes, and complex physics are still under work. For editorial planning that means preferring storyboards with calmer cameras and clear references, and not selling extreme sports or chaos physics as the default use case. The November 5, 2026 blog remains primary; developer docs and Cloud pricing add resolutions and quotas.
Separating It From Other AI Updates This Week
Gemini Omni 1.2 Flash is video productization. In parallel, code404zero covers cyber and coding releases — such as GPT-6 Cyber and Images 2.6 — that serve other surfaces. Avoiding cannibalization means keeping the focus keyword and H1 model- and spec-specific, not generic “AI video” titles without a version anchor. Internal links to Images 2.6, Pierre, and Astra keep the reading path in the same language.
Restated briefly: 15 seconds, multi-turn, −30% price, GA on API/Cloud, SynthID. No invented quality percentages, no price-chart or market-share speculation. Buyers read blog + docs + contract.
