On 5 November 2026 Anish Nangia, a product manager at Google DeepMind, wrote that Gemini Omni 1.2 Flash was rolling out that day. The model id is gemini-omni-1.2-flash. Native clips now run up to 15 seconds, up from 10 on Omni 1.1. The same post prices 720p video at $0.07 a second and says Google Vids can generate those clips at no additional cost. Those are two different bills.
What did the 5 November post ship?
The post is Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips, dated Nov 05, 2026. The opening line says that on that day Google was rolling out Gemini Omni 1.2 Flash and making it generally available across the Gemini API and Gemini for business. It introduces native 15-second video creation, multi-turn conversational editing, and enhanced reference following. It says the price dropped 30% from Gemini Omni 1.1, available at $0.07 per second for 720p video.
The post names no country. Duration can be set in one-second steps from 3 to 15 seconds. The workflow it describes is to draft at 360p or 720p, fix a detail with a follow-up edit instead of regenerating the whole clip, then upscale the best takes to 1080p or 4K.
What does the API table actually bill?
The pricing page linked from that post is Gemini Developer API pricing. The Gemini Omni 1.2 Flash section names the model id gemini-omni-1.2-flash and says it is generally available to developers on the paid tier. Free tier: Not available. Paid input is $1.50 per 1 million tokens for text, image, or video. Text output, including thinking tokens, is $9.00 per 1 million. Video output is $12.00 per 1 million.
The footnote is the $0.07 line. Billing uses 5,792 tokens per second of 720p. At $12.00 per 1 million tokens, the page says that is an effective price of approximately $0.07 per second, about $0.70 for a 10-second clip and about $1.04 for a 15-second clip. The same footnote prints 360p at $0.023 a second, at 1,931 tokens a second; 1080p with upsampling at $0.104 a second, at 8,688 tokens a second; and 4K with upsampling at $0.209 a second, at 17,376 tokens a second. Those figures are the page's own effective prices. They are not a Jakarta invoice.
Input tokens are a separate row. A reference image, a product shot, or a short motion clip is not included in the video-output footnote. Repeating $0.07 and skipping the input line understates the call.
Gemini Omni 1.1 Flash is still on the same page. Its video output is $17.50 per 1 million tokens. The footnote there says approximately $0.10 per second of 720p, using the same 5,792 tokens a second. The blog's 30% is that step, from about $0.10 to about $0.07. This article does not invent a sharper 1.1 per-second figure than the page's "approximately $0.10."
Effective December 3, 2026, the page says requests to gemini-omni-1.1-flash route to gemini-omni-1.2-flash and bill at 1.2 rates. The 1.1 price row is still printed. The Try link under that heading already opens gemini-omni-1.2-flash. Paid use of 1.2 is marked not used to improve Google's products. The free-tier cell says Not available. There is no free bucket on this table. Argon prices on the same page, input moving from $2.00 to $4.00 on 1 December 2026, are a different model, already covered in Gemini 4 Argon opens on Pro and Ultra, and the free app is off that list.
Where does "no additional cost" apply?
Developers are pointed at Google AI Studio and the pricing page. That door is the paid table. Free tier for this model is not available. Businesses are pointed at Google Cloud. The post says Gemini Omni is generally available there, that Provisioned Throughput is available, and that outputs carry an imperceptible SynthID watermark. It does not print a Provisioned Throughput price. The open detector is a different page: The SynthID Detector is now open to everyone.
The app list is separate. Flow is for directing scenes and multi-turn edits. Google Vids is where the post says 15-second clips cost nothing extra with a Google account or a Google Workspace subscription. That sentence is not the API's $0.07, and the post does not name the Workspace editions. The Gemini app line is a rollout to Google AI Ultra subscribers, not the free chat. Google Ads asset studio is named for scaling video campaigns. The post does not say a campaign in Indonesia is on. A paused photo upload for a custom avatar is a different Workspace Updates post. This 5 November post does not mention that upload.
What can the brief actually steer?
The post says generation can be guided with character, product, or style images, short motion clips up to three seconds, and audio cues, and that Omni 1.2 Flash holds those details from start to finish. It says faces, garments, packaging typography, and brand colors stay consistent across camera angles, so existing product photography or a character sheet can become a virtual try-on, a catalog spin, or a branded video ad. That is the product claim. It is not a measurement from a P2P Labs shoot.
The embedded demos are captions, not an operator log. The post shows a 15-second try-on from one model photo plus front-and-back shots of three outfits, a poster compared with Omni 1.1, background swaps, lip-sync from one scene plus four short voice references, and an FPV pass from two architectural photos. A later caption says a first-to-last push still changes small things: cats on the roof, and a soup ladle turning into chopsticks. The post is showing drift.
Speech is described as clearer in Flow, in Gemini for business, and in Google Vids. The post prints no language list. Indonesian on an older Vids avatar Help page is a spoken-script language, not a line on this post, and not a Jakarta rollout. Development is still working on fast sports action, rapid script changes, and complex physics. The Stan quote is about 15-second festival promos for an audience in India. India is not Southeast Asia. This article does not borrow partner quotes as local results.
What should a studio check before calling it live?
Read the model id before the brief goes out. gemini-omni-1.2-flash on the paid API is the $1.50 input, $9.00 text output, and $12.00 video output table. The 720p footnote is about $0.07 a second, about $1.04 for 15 seconds at 720p, before input tokens and before a 1080p or 4K upsample. A 4K upsample is the page's $0.209 a second, not the $0.07 line.
If the tool in front of the editor is Google Vids, the blog's line is no additional cost with a Google account or a Workspace subscription. Confirm the account, then confirm the clip control is actually on the project. Do not treat the API invoice and the Vids sentence as one number.
If the tool is the Gemini app, the post says 15-second video is rolling out to Google AI Ultra. A Pro subscription, or the free chat, is not that sentence. If a workflow still calls gemini-omni-1.1-flash, the pricing page says that on 3 December 2026 those requests route to 1.2 and bill at 1.2 rates. Until that date the 1.1 row is still printed at about $0.10 a second of 720p.
Do not read a demo caption, a partner quote, or the absence of a country name as availability in Indonesia. The post's own limit list is sports action, rapid script changes, and complex physics. The watermark says a SynthID signal may be in the file. It does not say which ad, which account, or which market bought the clip.