Gemini Omni Flash 1.2 Spotted: Release Window and Leaks

What has been spotted about Gemini Omni Flash 1.2?

Gemini Omni Flash 1.2, also written Gemini Omni 1.2 Flash, is an unreleased Google video generation model. @testingcatalog, which tracks AI releases, says model references were spotted on GCP and expects a minor upgrade over the current Omni 1.1 Flash.

Three evidence cards representing a cloud sighting, a price claim and a video clip

The model ID in the sighting is gemini-omni-1.2-flash. Google has not confirmed the model.

Separately, @chetaslua, a leaker, posted a clip attributed to the new model on November 5, 2026. The post calls it "spotted for public launch" and a "minor update over flash 1.1." The clip reportedly has the model explain its own pricing using Dragon Ball characters, which is not a pricing source.

OrcaRouter, a third-party blog, describes the public record as thin: a Google Cloud quota page screenshot and a ten-second video. It attributes the screenshot to @mrfanduu and says TestingCatalog folded the sighting into its daily brief the same day. None of these accounts is Google, and none of them includes a model card, price table, spec sheet or date.

A quota or service listing is a billing surface, not a product surface. It suggests the ID exists inside Google Cloud. It does not show that the model is ready for public use.

Gemini Omni Flash 1.2 release date and availability

No release date exists. Google's model page, developer guide and changelog still document only gemini-omni-1.1-flash, according to a third-party API blog that checked on October 28, 2026.

That same blog cites a community post from October 28 saying the model is deployed and releasing soon. Deployment and public launch are separate milestones, and the post gives no date.

Some API resellers already list a 1.2 page. One lists 0 active routes for 1.2 and says launch pricing is not announced. Another describes a 1.2 page with video controls for text, image references, first and last frames, or a source clip, at 360p to 4K. A listing like that is not an availability signal from Google, and neither page shows a working 1.2 route.

The timeline so far:

• October 28: a community post says 1.2 Flash is deployed and coming soon, per APIMaster.
• November 5: @chetaslua posts a clip attributed to 1.2, and @testingcatalog reports GCP references.
• Not yet: any Google announcement, documentation page or changelog entry.

Gemini Omni 1.2 Flash vs 1.1 Flash: what is known

Gemini Omni 1.1 Flash is Google's released Omni video model, and 1.2 is expected to be a small step from it. Every 1.2 detail below is unannounced. The 1.1 column comes from third-party summaries of Google's documentation.

One site publishes a side-by-side that labels every 1.2 entry a forecast, and it describes its 1.2 samples as illustrative. Do not read those as test results.

What Gemini Omni 1.1 Flash already does

Based on third-party summaries of Google's documentation, 1.1 Flash covers video generation and editing with native audio. It supports conversational edits that change one requested element while keeping the rest of the shot, plus video extension and first/last-frame interpolation.

The input side is broad. A reseller page lists text, image, audio and video references, with text-to-video, image-to-video, reference-to-video, edit and extend modes. Output runs 3 to 10 seconds, in landscape or portrait.

If 1.2 is a minor update, this is the baseline it will be judged against. A leaker's "minor" label points to the same task list with better quality or lower cost, but that is a read, not a spec.

Gemini Omni Flash 1.2 pricing: what is claimed

Official pricing for 1.2 is not announced. The only number in circulation is a claim from @mrfanduu, relayed by OrcaRouter, that 1.2 runs roughly 30% cheaper than 1.1 Flash.

That claim is a bare percentage. It has no rate, currency, tier or unit behind it, and none of the sources says Google has confirmed it.

For reference, a reseller lists 1.1 Flash from $0.055001 per output second at 720p, against a $0.10136 reference rate attributed to Google. The same reseller's base-route range runs from $0.055001 to $0.096801 per second. Reseller prices vary by route and are snapshots from October 28.

If the 30% figure held against the $0.10136 reference, the result would land near $0.071 per second. That is arithmetic on an unverified claim, not a quoted price.

How to prepare for Gemini Omni Flash 1.2 in ad workflows

Keep using 1.1 for now, and set up your tests so 1.2 can be compared the day it ships. Short clips with native audio, edits that change one element while keeping the rest of the shot, and first/last-frame control all suit product b-roll and hook variations.

A desk with a product bottle and a monitor showing abstract video timeline blocks

A practical prep list:

• Build a fixed test brief: one product, two supplied frames, 8 seconds, 16:9, so 1.1 and 1.2 can be compared on the same inputs.
• Check label and logo stability, since product geometry is where video models usually slip.
• Test one edit pass, such as changing a background color or prop while the product stays untouched.
• Hold budget decisions until Google publishes a rate. A leaked 30% figure is too weak to base a media or production budget on.

For ad teams, the useful questions are about control, not headline quality. Does a first/last-frame clip land on the supplied end frame? Does an extended clip join cleanly to its input? Does the pack shot hold across edits? Those checks apply to any new version.

You can run current Omni models next to other video models in the AdGraam AI Playground, and chain generation into repeatable workflows with Flow Builder. For hook testing methods, see the guide to hook rate and hold rate. For a comparable recent launch, see Kling 4 Flash.

What to watch next

The signals to watch are Google's model page, the Gemini API changelog and the Vertex AI catalog. Any of them listing gemini-omni-1.2-flash would turn this from a rumor into a launch.

Open questions: whether 1.2 changes clip length or resolution, what it costs per second, and whether upscaled 1080p and 4K tiers stay. A minor upgrade may mean the same controls with better consistency, but that is the leakers' read, not a Google statement.

Also watch whether API providers open real 1.2 routes. Today the pages that exist show 0 active routes and no pricing, so a live route with a posted rate would be the first hard sign of availability.