Google extends AI video generation to 15 seconds and cuts 720p rate by 30%

Google released Gemini Omni 1.2 Flash on 5 November, extending the maximum length of an AI-generated clip in a single generation from 10 to 15 seconds and cutting the 720p price by 30 per cent to $0.07 per second. The model is the latest increment in a fast-refreshed product line: the Omni family was presented at Google I/O on 19 May, and its predecessor, 1.1 Flash, arrived on 27 August. The update targets commercial video work — product explainers and social advertising — where clip length and per-second cost decide whether generation is economic at scale.

The mechanism is a lengthened single pass plus cheaper iteration. Users specify any duration between 3 and 15 seconds in one-second increments, rather than stitching extensions onto an existing scene. Google advises drafting at 360p or 720p, correcting details by conversational instruction instead of regenerating, and upscaling only the chosen take to 1080p or 4K — the latter billed at about $0.21 per second, roughly three times the 720p rate, according to Frandroid. Reference inputs — character or product images, video clips of up to three seconds and audio cues — are meant to keep faces, clothing, packaging text and brand colours stable as the camera angle changes, and multi-turn editing allows backgrounds, lighting, weather or aspect ratio to be altered without restarting.

Speech and camera work received attention too: the model separates speakers in multi-person dialogue and syncs lip movement, and sustains continuous moves such as first-person drone flight or a long zoom-out with fewer unwanted cuts. The caveats are Google's own. It acknowledges remaining difficulty with fast sporting action, abrupt script changes and complex physics, and it has published no numerical comparison against rival models; the claimed improvements rest on demonstrations the company selected. Generated video carries the invisible SynthID watermark.

Deployment is staged and uneven across services. In the Gemini app, 15-second clips are reaching Google AI Ultra subscribers progressively. Google Vids offers the longer clips without surcharge on a Google account or Workspace subscription, according to Frandroid, while AI Times reported a free allowance there of up to three 15-second videos a month and Liberty Times said free users could try the feature in both Google Vids and Google Ads. Developers can test the model in Google AI Studio through the Gemini API, and enterprises reach it via the Google Cloud console, including provisioned throughput. The milestone to watch is the completion of that staged rollout to Ultra subscribers, with no date disclosed.