Google: Gemini Omni 1.2 Flash

Google moved gemini-omni-1.2-flash to general availability on the Gemini API on 5 November 2026, offering video clips up to 15 seconds with native audio, multi-turn editing through the Interactions API, and a video output price cut from $17.50 to $12.00 per million tokens. Inputs include up to 6 reference images and short video references of up to 3 seconds each, with MP4 output from 3 to 15 seconds at 360p through 4K. The model code gemini-omni-1.2-flash is available to paid-tier Gemini API users, with the free tier unsupported, and it also reaches Google Cloud, Flow, Google Vids, and the Gemini app. Drafting at 360p costs about $0.023 per second before upscaling, and the older Omni 1.1 model name is slated to silently reroute to 1.2 on 3 December 2026, so callers should update pinned model identifiers.

Google announced Gemini Omni 1.2 Flash is generally available across the Gemini API and Gemini for business on Google Cloud, in a launch post authored by Anish Nangia, Product Manager at Google DeepMind. It is the first time the Gemini Omni line has reached general availability, and outputs carry an imperceptible SynthID watermark consistent with Google's AI-content labeling approach. The headline upgrade expands native video length from 10 seconds to 15 seconds per generation, with duration selectable in one-second increments between 3 and 15 seconds. Google frames the longer window as sufficient for a complete product explainer, social ad, or branded scene without chaining shorter clips, and the model is accessible via AI Studio, the Gemini API, the Google Cloud console, Flow, Google Vids, Google Ads, and the Gemini app for Ultra subscribers.

Google is rolling out Gemini Omni 1.2 Flash, extending generated video to up to 15 seconds and cutting the 720p price to about $0.07 per second, roughly 30 percent below Gemini Omni 1.1. The model reaches general availability through the Gemini API and Gemini for business, with maximum clip length rising from 10 seconds on Omni 1.1 to a selectable 3-to-15 seconds in one-second steps for both native generation and editing. Users can draft at 360p or 720p and upscale selected results to 1080p or 4K, while conversational editing applies successive changes through follow-up prompts and can lower workflow cost versus full regeneration. References may combine character or product images with audio cues and motion clips up to three seconds, which Google says preserves faces, clothing, packaging text, and brand colors across camera angle changes.

Google announced Gemini Omni 1.2 Flash on November 5, 2026, making the video generation and editing model generally available. The update raises native clip length to 15 seconds, adds multi-turn conversational editing, and cuts 720p pricing by 30 percent to roughly $0.07 per second of video. The model produces synchronized video and native audio from text, image, and video inputs, and the longer runtime is aimed at fitting complete stories, product explainers, and full social ad spots into one generation. Developer documentation supports whole-second durations from 3 to 15 seconds, up from a 10-second maximum in Omni 1.1, with 10 seconds as the default. Output defaults to 720p and 16:9, teams can draft at 360p or 720p and upscale to 1080p or 4K, and conversational editing can swap products or backgrounds, change lighting, or switch aspect ratio while leaving unmentioned elements intact. A previous-interaction identifier preserves context between API requests, and Gemini Omni 1.1 Flash stays available during a four-week transition before requests route to 1.2 automatically on December 3, 2026.

DataCamp hands-on testing of Gemini Omni 1.2 Flash found that generated and edited clips run 3 to 15 seconds in one-second increments, up from Omni 1.1 Flash's 10-second ceiling, with video and sound produced together and an invisible SynthID watermark embedded in each output. First-and-last-frame generation is limited to 10 seconds, and extension caps differ by surface: the Gemini API caps finished video at 30 seconds while Google Cloud documents up to 40 seconds. In the author's marble-run experiment, one clip returned as a continuous 15-second 720p shot, but the marble briefly vanished and reappeared behind the dominoes. A color edit changed only the marble while preserving original frame timing, and a stop-motion restyle altered textures but kept smooth motion, though the audio drifted slightly with each successive edit. The piece also notes the model accepts text, image, and video inputs and returns video with sound, unlike Gemini text models.

Google DeepMind published the model card for Gemini Omni 1.2 Flash on 5 November 2026, describing it as a step toward models that can create and edit anything from any input, starting with video. It combines Gemini's intelligence with Google's generative media models, positioning the release as a leap forward in world understanding, multimodality, and editing. The card lists inputs as text strings, images, audio, and video files, with outputs being high-resolution video with audio, and states that conversational editing provides a more natural way to change generated footage. Architecture, training data, and hardware details are delegated to the Gemini Omni Flash model card, and distribution covers the Gemini App, Gemini Enterprise App, and Google Cloud channels.