On this page
Google published a product blog titled Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips, with datePublished 2026-11-05T17:00:00+00:00, which is 2026-11-06 01:00:00 Asia/Shanghai. The post is bylined to Anish Nangia, Product Manager at Google DeepMind. It says Google is rolling out Gemini Omni 1.2 Flash and making it generally available across the Gemini API and Gemini for business. The same post describes native video generation and editing up to 15 seconds, multi-turn conversational editing, and a self-reported 30% price drop from Gemini Omni 1.1, to $0.07 per second for 720p video.
At 01:08:32 Asia/Shanghai (2026-11-05T17:08:32.000Z), the Google AI Studio public account posted that the model is now available in AI Studio and the Gemini API, listing faster generation and 30% lower cost, longer clips, editing in conversation, more natural speech, and more controllable shots. An attached video is Google's own demo, not a sample generated here. Google DeepMind later forwarded the same post rather than publishing a separate product note.
OpenRouter's public catalog then listed google/gemini-omni-1.2-flash. The models API created stamp is 1793898961, which is 2026-11-06 01:16:01 Asia/Shanghai (17:16:01 UTC). The listing name is Google: Gemini Omni 1.2 Flash, with a context length of 131,072 tokens.
Later the same morning, at 02:12:57 Asia/Shanghai (2026-11-05T18:12:57.000Z), @GeminiApp said Gemini Omni 1.2 Flash is rolling out starting today to all Google AI subscribers in the Gemini app, and that Google AI Ultra subscribers can now create and edit videos up to 15 seconds.
We haven't tested generation quality, timing, or billing on AI Studio, the Gemini API, or OpenRouter, and we haven't logged into the Gemini app to check who already has access. The notes below follow Google's product blog, the public posts, the developer documentation and pricing page, and OpenRouter's official model page plus the public models list.
Clip length, editing, and native audio
Developer documentation now presents gemini-omni-1.2-flash as the Gemini API model for video generation and conversational video editing. Output video goes up to 15 seconds with native audio, longer than Omni 1.1's 10 seconds. Duration can be set in whole seconds from 3 to 15; if unset, the default is 10 seconds. First-and-last-frame interpolation and each extension turn support 3 to 10 seconds, with extended videos allowed to reach 30 seconds in total. Aspect ratios are 16:9 and 9:16. The default resolution is 720p, with 360p, 1080p, and 4K also listed; the page describes 4K as upscaled. Character consistency is described as supporting up to 6 reference images.
The blog and the developer documentation both highlight multi-turn conversational editing. On Google's Interactions API, that chain uses previous_interaction_id. OpenRouter's official model page describes the same kind of edit or extension through previous_job_id. Those are two different parameter names on two different interfaces, not one shared argument.
The product blog describes drafting at 360p or 720p, fixing a detail with a follow-up edit, then upscaling a chosen result to 1080p or 4K. It also says outputs carry an imperceptible SynthID watermark; how Google marks AI-generated media is covered in Gemini Lets You Check if Media Was Made with Google AI. The same blog mentions Flow, Google Vids, Google Ads, and creating 15-second videos in the Gemini app (rolling out to Google AI Ultra subscribers); those mentions come from the product blog; those apps were not logged into again to re-check them.
Gemini app rollout
At 02:12:57 Asia/Shanghai (2026-11-05T18:12:57.000Z), @GeminiApp said the model is rolling out starting today to all Google AI subscribers, with multi-turn conversational edits and sharper visual stability in the Gemini app. The same post also said Google AI Ultra subscribers can now create and edit videos up to 15 seconds. It quoted an earlier @Google introduction post; the attached video there is Google's demo, not a sample generated here. The blog's Ultra 15-second Gemini-app line and this all-subscribers rollout plus Ultra create-and-edit ceiling are two separate statements; the Gemini app was not logged into to re-check who already sees which capability.
This launch is about generating and editing video. A separate note, How to Choose Gemini Video Mode—Static 1 FPS or Agentic Retrieval, is about understanding existing video, not about Omni's generation model.
What the official ~30% lower video price refers to
Google's pricing page lists Gemini Omni 1.2 Flash as generally available on the paid tier of the Gemini API. The Free Tier column is Not available. Paid prices per 1 million tokens (USD) are input $1.50 (text / image / video) and output $9.00 (text) / $12.00 (video), including thinking tokens.
A footnote says billing uses output tokens at 5,792 tokens per second of 720p video. At $12.00 per 1 million video tokens, that is about $0.07 per second. The same footnote lists about $0.023 per second for 360p, about $0.104 per second for 1080p with upsampling, and about $0.209 per second for 4K with upsampling.
Gemini Omni 1.1 Flash is still listed as the previous generation. Its paid input price is also $1.50, and its text output price is also $9.00. Its video output price is $17.50 per 1 million tokens, or about $0.10 per second at 720p. The official "about 30% lower" claim is about video pricing—the blog's $0.07-per-second 720p figure, and the table's video output row from $17.50 to $12.00. Input did not drop 30%, and neither did text output. The $9 text-output figure is not the video rate.
Those numbers come from the pricing page and the product blog; we have not reconciled a sample invoice against them. For the broader Gemini Developer API wall, see Gemini API Pricing (November 2026).
The OpenRouter listing
OpenRouter's official model page for google/gemini-omni-1.2-flash describes the same Google model and says video generation is supported only through the Video Generation API. The models API records a context length of 131,072 and max_completion_tokens of 65,536. The FAQ dates the release as November 5, 2026.
The page header shows $1.50 / $9 per 1M, matching pricing.prompt 0.0000015 and pricing.completion 0.000009, and matching Google's input and text output rates. That header pair is not the full cost of generating video. The FAQ adds separate per-second rates for video with audio, from $0.02317 per second at 360p to $0.2085 per second at 4K. A provider table may display about $0.023172 per second—that is the 360p-end figure, not the only video unit price. Google's pricing page uses $12.00 per 1 million video output tokens; the two sites should be read side by side rather than averaged into one homemade output price.
The same page notes that the listing is not currently available through EU in-region routing. The catalog already included google/gemini-omni-1.1-flash; what follows is the 1.2 listing around 01:16. A same-day OpenRouter routing change, on a different topic, is covered in OpenRouter Adds Auto Code and Moves Auto to Jev Router.
Veo 3.1 Preview on the Gemini API, and the Omni 1.1 window
The Gemini API video overview says the Veo 3.1 Preview in the Gemini API has been shut down. A warning names three Preview IDs—veo-3.1-generate-preview, veo-3.1-fast-generate-preview, and veo-3.1-lite-generate-preview—and says they were shut down on October 22, 2026. The suggested paths are Gemini Omni Flash (gemini-omni-1.2-flash), or the GA models still offered on the Gemini Enterprise Agent Platform. That is not a claim that every Veo model—or enterprise Veo GA—has been turned off. Veo and Gemini Omni are different models, and the shutdown date is 2026-10-22, not this announcement morning.
gemini-omni-1.1-flash remains available during a 4-week transition window. From December 3, 2026, requests to that ID are to be routed automatically to gemini-omni-1.2-flash, with no code changes required, and billed at the lower 1.2 rates.
What we haven't checked
We haven't generated clips in AI Studio, called gemini-omni-1.2-flash on the Gemini API, or run google/gemini-omni-1.2-flash on OpenRouter. There is no independent timing figure, invoice, or quality comparison against Omni 1.1. The figures above are published descriptions as of 2026-11-06 (Asia/Shanghai).
In short
On the morning of 2026-11-06 (Asia/Shanghai), Google introduced Gemini Omni 1.2 Flash as generally available: native video up to 15 seconds (Omni 1.1: 10 seconds), multi-turn conversational editing, native audio, and a self-reported 30% lower video price (about $0.07 per second at 720p). Paid list prices are $1.50 input, $9 text output, and $12 video output per 1 million tokens; the free tier is not available. OpenRouter listed google/gemini-omni-1.2-flash around 01:16 (context 131,072, header $1.50 / $9, plus per-second video rates). Around 02:12, @GeminiApp said it is rolling out starting today to all Google AI subscribers in the Gemini app, with Ultra able to create and edit up to 15 seconds. Three Veo 3.1 Preview IDs were shut down on the Gemini API on 2026-10-22; Omni 1.1 stays available until 2026-12-03. We haven't tested the new model yet.