# Gemini Omni 1.2 Flash Gemini Omni 1.2 Flash is the version of [[Gemini Omni]] that [[Google DeepMind]] released on 5 November 2026. It's a video generation and editing model, not a chat model (see the naming problem below). Google's pitch fits in the blog post title: "fast and cost-effective with longer clips". It's also the first time Gemini Omni is generally available, with Google Cloud's enterprise infrastructure behind it. ## What changed since Omni 1.1 - **15-second clips.** Up from 10 seconds with Omni 1.1. You choose the duration in one-second steps, from 3 to 15 seconds. That's enough for a full social ad, a product explainer, or a short scene in one generation - **Better reference adherence.** You can combine character, product or style images, short motion clips (up to three seconds) and audio cues. Faces, garments, packaging text and brand colors are supposed to hold from the first frame to the last. Google's example: one model photo plus front and back shots of three outfits, turned into a 15-second virtual try-on with 360-degree turns - **Less drift.** Characters, props and environments morph less across motion and camera moves, including when you interpolate between a first and a last keyframe or extend a clip - **Multi-turn editing on full clips.** Swap a product or a background, change the lighting, the weather or the aspect ratio, turn after turn, without generating from scratch. This is how you'd produce ad variants for different audiences - **More natural speech.** Better lip-sync and distinct voices when several characters talk - **Camera control.** Long continuous moves (FPV flythroughs, crash zoom-outs, orbits) with fewer unwanted jump cuts Google is upfront about what still fails: fast sports action, rapid script changes and complex physics. That matches what testers said about the first Gemini Omni (objects that disappear or merge under physical stress). ## Pricing and availability Through the API, it's **$0.07 per second of 720p video**, 30% less than Omni 1.1. A full 15-second clip costs about $1.05. Google suggests a workflow to keep costs down: draft at 360p or 720p, fix details with a follow-up edit instead of a new generation, then upscale the best result to 1080p or 4K. Where you can use it: - The Gemini API (prototype in [[Google AI Studio]]) and the Google Cloud console, with Provisioned Throughput - Gemini for business - Flow, for scenes and multi-turn edits - Google Vids, where 15-second clips come at no additional cost Every output carries an invisible SynthID watermark. The price drop only holds for the API. In the Hacker News thread, users reported the opposite in Flow: 18 credits for a six-second video with Omni 1.2, against 10 credits with Omni 1.1. One commenter summarized it as "double the cost in Flow, but 30% cheaper via the API", adding that Flow had been 5-6x cheaper than the API until now. These are user reports, not official numbers. ## How it was received The [[Hacker News]] thread (339 points, 231 comments) barely discussed the model. Most of it was about how hard it is to pay Google for AI: - **Access and billing.** Billing errors, identity verification requests, and a Google Cloud console that people find hard to navigate. Several paying subscribers said they couldn't try the models at all. One wrote: "I literally don't care about their models anymore because I can't even try them without paying for them?" - **Plans.** Users couldn't tell which plan includes which model, or whether they need a personal or a work account. Workspace customers described having to upgrade the whole organization to give AI to one person: "Upgrade the entire org, or nobody gets AI" - **AI Studio.** "AI Studio is legitimately the worst UX experience on the internet" was one of the most upvoted lines - **Rollout.** An Ultra subscriber only saw Veo and Omni 1.1; someone else had Omni 1.2 in AI Studio on a normal account; others said Google Vids would take 1-3 days - **Naming.** "The 'omni' naming for a video-only model is so confusing." Another commenter had to explain that this is "a new video model", not a replacement for [[Gemini 4 Argon|Gemini 4]] - **The counterpoint.** Some commenters said the models themselves "just work" in their workflows, and recommended going through third-party routers like [[OpenRouter]] to avoid Google's billing altogether. A few suspected that the criticism was overblown Commenters blamed it on organizational silos (Vertex, DeepMind and the Gemini app run by different teams) and on a company that rarely talks to its customers. I can't verify that, but the pattern of complaints is consistent across the thread. ## My take The model update is solid and practical. Fifteen seconds, consistent references and edits that keep the rest of the clip intact are what you need for real ad and product work. The API price makes iteration affordable: about a dollar per full clip. But the thread says more about Google than about the model. When the top comments on a launch are about billing pages and plan confusion, the product isn't the bottleneck anymore; access is. Google keeps shipping competitive models (this one, [[Gemini 4 Argon]], [[Nano Banana]]), and keeps making them hard to buy. If you want to try Omni 1.2 Flash, I'd start with the API and AI Studio, not with Flow: the credit math there went the wrong way. Also note that this is a video model. If you only need a short loop for social media, an image model that makes GIFs, like [[ChatGPT Images 2.6]], is a cheaper way to iterate. ## References - Announcement (2026-11-05): https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/ - Hacker News discussion: https://news.ycombinator.com/item?id=50392118 ## Related - [[Gemini Omni]] - [[Gemini]] - [[Google DeepMind]] - [[Google AI Studio]] - [[Vertex AI]] - [[Veo 3]] - [[Nano Banana]] - [[Genie 4]] - [[Sora]] - [[ChatGPT Images 2.6]] - [[Gemini 4 Argon]] - [[OpenRouter]]