Fifteen seconds, seven cents a second

The Google post Gemini Omni 1.2 Flash: 15-second video and lower pricing carries the timestamp 2026-11-05T17:00:00+00:00, which is 01:00 in Beijing on Nov. 6, inside this morning window. The byline is Anish Nangia, a product manager at Google DeepMind. The body says Gemini Omni 1.2 Flash is generally available today on the Gemini API and in Gemini for business. A summary higher on the page is labeled as generated by Google AI and still experimental. A channel that appears only in that summary is not used here.

The body says the price is 30 percent lower than Gemini Omni 1.1, at $0.07 per second for 720p video. Beyond the price, the company describes a way to hold cost down: draft at 360p or 720p, change a detail with a follow-up edit instead of regenerating the whole clip, then upscale a chosen result to 1080p or 4K. That is a product description, not an independently measured cost table.

Native video moves from 10 seconds on Omni 1.1 to 15 seconds. Duration can be set in one-second steps from 3 seconds to 15. A reference can combine character, product, or style images, a motion clip of up to three seconds, and audio cues. The company says faces, garments, packaging type, and brand colors hold across camera angles. The virtual try-on passage is the blog's own demonstration, with no sample size and no error bar. Multi-turn editing covers a full 15-second clip and is described as a way to change the product, background, light, weather, or frame. The page does not print a score against another lab's model.

The try-on passage is more specific: one model photo plus front and back shots of three outfits, then a 15-second try-on across 360-degree turns and match cuts, with the model's identity and each garment's cut, drape, and back details held. That is still the blog describing one demonstration, with no sample size. The poster comparison is qualitative. The blog says that, given the same static exhibition poster and a five-layer animation prompt, Omni 1.1 Flash does not match the reference as closely as 1.2. It names materials. It does not give an error. The speech passage says dialogue is more fluid and lip-sync is clearer, and it lists Flow, Gemini for business, and Google Vids as places where that shows up. Lip-sync has no error rate.

What can be fixed is the product account the blog published at 01:00 in Beijing. The $0.07, the 30 percent, and the 15 seconds are in the body. Consistency is the company's statement. The image is the post's announcement graphic. Sample frames sit around the title. It is not a score table.