Gemini Omni 1.2 Flash Is Out: 15-Second Videos and 30% Cheaper Video Output

By Lesia Kapliuk · Published November 5, 2026 · Updated November 5, 2026

Update (Nov 5, 2026, evening): Google has launched Gemini Omni 1.2 Flash. Between 17:03 and 18:03 UTC the changelog, pricing and rate limits pages were updated, and Google published an announcement. By 18:42 UTC the model was in the AI Studio Playground on our AI Pro account, marked New and Paid. In short: up to 15-second videos, video output at $12.00 per 1M tokens instead of $17.50, doubled paid rate limits, and from December 3, 2026 requests to Omni 1.1 go to 1.2. We also generated a 15-second test clip on our AI Pro account; in AI Studio without an API key it was free. Full figures and the clip are in the launch section below. We changed the title from "Gemini Omni 1.2 Flash Is Coming" now that it has launched; the original pre-launch text is unchanged, with short update notes where it is out of date.

On this page: price, limits, 1.1 vs 1.2 · our 15-second test · what we saw before the launch

Gemini Omni 1.2 Flash at launch: price, rate limits, 1.1 vs 1.2

Added on November 5, 2026, after Google's announcement. Source: the Gemini API changelog, pricing and rate limits pages.

	Omni 1.1 Flash	Omni 1.2 Flash
Model code	`gemini-omni-1.1-flash`	`gemini-omni-1.2-flash`
Status	GA since August 27, 2026	GA since November 5, 2026
Free tier	not available	not available
Clip length	3–10 s	up to 15 s
First and last frame	supported	3–10 s
Extension	supported	3–10 s per turn, up to 30 s in total
Reference images	supported	up to 6
Input price, per 1M tokens	$1.50	$1.50
Text output, per 1M tokens	$9.00	$9.00
Video output, per 1M tokens	$17.50	$12.00

Gemini Omni 1.2 Flash pricing

Video output costs $12.00 per 1M tokens, which Google calls 30% lower than 1.1. Billing counts output tokens per second of video: 5,792 tokens per second at 720p. Google's own examples work out to about $0.07 per second, ~$0.70 for a 10-second clip and ~$1.04 for a 15-second clip. Other resolutions: 360p about $0.023 per second, 1080p with upsampling about $0.104, 4K with upsampling about $0.209.

Gemini Omni 1.2 Flash rate limits

Google doubled the paid-tier limits. Standard interactive limits for gemini-omni-1.2-flash:

Usage tier	Requests per minute	Requests per day
Free	not supported	not supported
Tier 1	8	40
Tier 2	16	200
Tier 3	40	2,000

What happens to Omni 1.1 on December 3, 2026

From December 3, 2026, all API requests to gemini-omni-1.1-flash are routed automatically to gemini-omni-1.2-flash and billed at 1.2 rates. You don't need to change code to keep working, but output will come from the new model. Since 1.2's video output is cheaper, the bill for the same video length should go down; test your prompts on 1.2 before the switch if the look of your videos matters.

We tried it: a 15-second clip on an AI Pro account

On the evening of the launch we generated one video in the AI Studio Playground on our Google AI Pro account. Settings: 16:9, 15 seconds, 720p, 24 fps, video task "Unspecified". As the reference image we attached our own logo (a dark tile with a white eye and an amber cursor) and asked for three shots: the logo on a desk next to a laptop, a push-in while the laptop screen scrolls, and a close-up. The full prompt is in the screenshot below.

Free in AI Studio. AI Studio did not ask for a billing project or an API key, and its token panel says: "Usage on AI Studio is free when no API key is selected." We don't know yet how many clips a day the AI Pro allowance covers.
Time: 49.4 seconds from Run to the finished video (the model first shows a collapsed "Thoughts" block), then 2.9 seconds until the reply with the embedded video appeared.
Output: exactly what we asked for: an MP4 of 15.04 seconds, 1280×720, 24 fps, with a stereo audio track.
Tokens: 88,927 in total, out of a 131,072-token budget shown in AI Studio: 1,252 input (1,089 for the reference image, 163 for the text), 86,880 video output and 795 thought tokens. The video figure is exactly Google's billing rule of 5,792 tokens per second of 720p video (15 × 5,792).
Cost estimate doesn't match the price list. AI Studio estimates what the same request would cost through the API: $0.0019 input plus $0.7819 output, $0.78 in total. $0.7819 is 86,880 tokens at $9.00 per 1M, the text output rate. At the $12.00 video output price on the pricing page, the same tokens come to about $1.04, which is also Google's own example for a 15-second clip. The estimate also leaves out the 795 thought tokens, although the price list says output price includes thinking tokens. We trust the pricing page over the estimate; one of the two will probably be corrected.
Consistency: the logo keeps its shape and colours in all three shots, and the cursor blinks. The shot order and timing follow the prompt. One thing it did not follow: we asked for "silence" at the end, but the audio track never drops below -45 dB, so there is no silent stretch.
AI Studio still shows "This model is not stable and may not be suitable for production use" in the sidebar for this model, even though the API docs call 1.2 generally available.

AI Studio now also shows a "Generate videos with Gemini Omni 1.2 Flash" block on our account, so the launch text we found earlier in its data has gone live in a new form.

Before the launch: what we saw first

The text below was published before Google's announcement and is unchanged.

TL;DR: Google's video model Gemini Omni 1.2 Flash (gemini-omni-1.2-flash) appeared in the public Gemini API model list on November 5, 2026 at 16:03 UTC. About half an hour later, the data Google AI Studio loads for its "now available" banners got a new entry with a launch time of 16:30 UTC: "Create 15 second videos with Gemini Omni 1.2 Flash". The banner is not shown on our account yet, but its text promises videos from 3 to 15 seconds with native audio, "stronger character and subject consistency" and 4K upscaling, for AI Pro, AI Ultra and paid API key users. On our AI Pro account the model is not available yet: at 16:45 UTC the Playground model picker and the rate-limit page still showed only Omni 1.1 Flash. The current model, Omni 1.1 Flash, makes 3–10 second clips. At the time of writing the documentation, the model page, the pricing page and the changelog still describe only 1.1, so price and limits for 1.2 are not published yet.

Update: price and limits are now published, see the launch section.

How we found it

We check the Gemini API's public models.list every hour with our own API key, and we save the pages Google AI Studio loads while we browse it under our own account. On November 5 the hourly check at 16:03 UTC found a new entry in the v1beta and v1alpha lists: gemini-omni-1.2-flash, display name "Gemini Omni 1.2 Flash". It is not in the stable v1 list.

AI Studio gets the texts of its "now available" banners from its own server, each with a launch time. The list our browser received at 16:27 UTC did not have Omni 1.2 yet. The one at 16:31 UTC did, with a launch time of 16:30 UTC and a link to a Playground chat with gemini-omni-1.2-flash. On our account the banner itself did not appear on screen, so we can't show it: we quote its text from that data. A ready-made launch text is a strong sign of a launch, but until Google announces it, it is not one.

We have not generated any videos with 1.2 for this article. Everything about what the model does comes from that banner text, not from our tests.

What's new compared with Omni 1.1 Flash

Gemini Omni Flash launched as gemini-omni-flash-preview on June 30, 2026 with 3–10 second videos at 720p. The GA version, gemini-omni-1.1-flash, followed on August 27 and added video extension and first-and-last-frame interpolation. Here is how 1.2 compares, going by what Google has published so far:

	Omni 1.1 Flash (docs)	Omni 1.2 Flash (banner text, not yet shown)
Model code	`gemini-omni-1.1-flash`	`gemini-omni-1.2-flash`
Clip length	3–10 s	3–15 s
Audio	generated with the video	"native audio"
Consistency	reference images and videos	"stronger character and subject consistency"
4K	upscaled output	"4K upscaling"
Who can use it	paid tier of the Gemini API	AI Pro, AI Ultra and paid API key users
Price	input $1.50, output $9.00, video output $17.50 (AI Studio picker)	not published

The longer clip is the clear change. Audio and 4K upscaling are already in 1.1, so we read those lines of the banner as a reminder rather than new features. "Stronger consistency" is Google's claim, and we have not measured it.

Update: the launch figures are in the table above.

The access line is worth noting. The pricing page describes 1.1 as "generally available to developers on the paid tier", and its free tier says "Not available". The 1.2 banner adds AI Pro and AI Ultra subscribers, which matches the Google AI plans page: those plans unlock paid models in the AI Studio Playground. So a Pro subscriber should be able to try 1.2 in AI Studio without an API key with billing. On our AI Pro account that is not the case yet: 15 minutes after the banner's launch time, the Playground listed only gemini-omni-1.1-flash (marked Paid), the rate-limit page showed limits only for 1.1, and the banner was not displayed. Our guess is a staged launch that hasn't reached our account; we'll update this article when 1.2 appears for us.

What the API list says, and what it doesn't

The API entry for 1.2 lists two methods, generateContent and countTokens, the same as 1.1. Batch and context caching are not listed for either.

The token limits in the list (131,072 in, 65,536 out) are the same for 1.1 and 1.2, but the 1.1 documentation gives a 1,048,576-token context window. The list and the docs disagree for 1.1, so we don't treat the list's numbers as the real limits of 1.2.

What we don't know

Price. The pricing page has only 1.1. We don't know if 1.2 costs the same.
Limits for subscribers. How many 1.2 videos AI Pro or AI Ultra subscribers get per day.
What happens to 1.1. When 1.1 went GA, Google set a deprecation date for the preview. Nothing like that is published for 1.1 yet.
Whether 15 seconds applies to extension and editing. In 1.1, input videos for editing and extension must be 10 seconds or less.
The Gemini app. The few Gemini app requests we saved that evening don't mention Omni at all, so we can't say whether 1.2 is coming there.

Update: price, the fate of 1.1 and extension length are now answered above. Daily quotas for AI Pro and AI Ultra subscribers in AI Studio, and the Gemini app, are still unknown.

Third-party sites already publish 1.2 "specs" and playgrounds. Treat them as unconfirmed until Google's own documentation lists them.

How to check if you have Gemini Omni 1.2 Flash

In AI Studio: open the model picker in the Playground and look for Gemini Omni 1.2 Flash, or check the rate-limit page for a 1.2 row. A "Create 15 second videos" banner would be another sign.
With the API: list the models with your key and look for gemini-omni-1.2-flash in v1beta:

curl "https://generativelanguage.googleapis.com/v1beta/models?pageSize=1000" -H "x-goog-api-key: $GEMINI_API_KEY"

Keep the key in an environment variable, not in the command line.