news

Gemini Omni 1.2 Flash goes GA at $0.07 a second for 720p, with 15-second clips and no published scores

Google cut the API price 30% against Gemini Omni 1.1 Flash and raised native generation from 10 seconds to 15, while publishing no benchmark number behind the quality claims.

By Priya Shenoy · 6 November 2026

Google made Gemini Omni 1.2 Flash generally available on November 5, 2026, with two changes that can be checked and one set of claims that cannot. The checkable parts: native video generation and editing now run to 15 seconds instead of 10, and the API price is $0.07 per second of 720p video, which Google states is 30% below Gemini Omni 1.1 Flash. The unverifiable part is everything about output quality, because neither the consumer announcement nor the Cloud post carries a score, a win rate or a named comparison.

Clip length becomes a dial, from 3 seconds to 15

The duration change is not just a higher ceiling. Google's developer account describes an increased native limit plus the ability to pick an exact length anywhere between 3 and 15 seconds, which matters more for ad and social formats than the headline number does. The previous generation topped out at 10 seconds.

Alongside the length change, Google lists multi-turn conversational editing, stronger adherence to reference assets, upgraded character speech, smoother shot continuity with fewer jump cuts, and support for camera moves such as flythroughs and zoom-outs.

The 30% cut is priced for one resolution only

The Cloud post gives a single explicit figure, $0.07 per second of 720p video, and describes the savings at other resolutions as similar rather than naming them. A customer quote in the same post refers to 15-second generations at 1080p, so longer clips are not restricted to the one resolution that has a stated price, but the per-second cost at 1080p is not in the announcement. Anyone budgeting above 720p is working from an adjective.

Our radar tracks what this model costs across the platforms that resell it:

Google names three weaknesses and zero numbers

The closest thing to a quality disclosure in the launch is a list of things that are still being worked on: fast sports action, rapid script changes and complex physics. That is more candid than most launch posts, and it is still not a measurement. There is no Elo, no human preference rate, no VBench-style breakdown, and no side-by-side against the 1.1 version that the price comparison is drawn against. The pricing claim is anchored to a specific predecessor; the quality claim is anchored to nothing.

The model inherits the slot Veo 3.1 vacated

Gemini Omni 1.1 Flash was the single named replacement when Google turned off the three Veo 3.1 preview endpoints, so developers who migrated off Veo 3.1 are now on their second Omni version in a matter of weeks. Availability for 1.2 Flash spans the Gemini API and Google AI Studio, the Google Cloud console with Provisioned Throughput, Google Vids, Google Flow and the Gemini app, with the longest clips in the app tied to the Ultra tier during rollout.

For a model being positioned for production video at scale, the asymmetry is the story. Google quantified the thing that costs it money and left the thing it is selling to adjectives.

Sources: Google, Google Cloud.

Priya Shenoy: Tracks what AI video actually costs across the platforms that resell access to the same handful of models. Treats a pricing page as a claim, not a fact, until someone checks it.