CVCOVANewsComputing, Opportunities, Value & Analysis
Artificial Intelligence4 min read

Gemini Omni 1.2 Flash: Google extends its AI videos to 15 seconds and cuts the price by 30%

Google DeepMind has made its new AI video model available, with longer clips, better fidelity to brand references and a lower rate. A 15-second clip at 720p costs about one dollar, but for creators and small businesses in Latin America the real cost is still unclear.

By COVA News newsroom
Gemini Omni 1.2 Flash: Google extends its AI videos to 15 seconds and cuts the price by 30%

For about one dollar, you can generate a 15-second AI commercial featuring the product, the brand's colors and even the typeface on the packaging. That is the pitch for Gemini Omni 1.2 Flash, the video model that Google DeepMind declared available on November 5, 2026, in the Gemini API, which developers use to connect their applications to Google's models, and in Google Cloud. According to the company, clips go from 10 to 15 seconds and the per-second price drops 30%, to US$0.07 at 720p resolution.

The announcement spread quickly. The Decoder covered it the same day, it appeared on Techmeme, and the Hacker News thread, on the tech community forum, drew 189 points and 146 comments. In that conversation, however, frustration with the access, naming and pricing of Google's AI products weighed more heavily than analysis of the model itself.

More seconds, more brand control

The Omni Flash family was unveiled at Google I/O in May with a cap of 10 seconds per clip, the same limit version 1.1 kept. With 1.2, Google is targeting advertising and commercial content. According to Unite.ai, the model accepts several references at once: images of a character, a product or a style, motion clips of up to 3 seconds, and audio tracks. Google says faces, clothing, packaging and brand colors stay consistent even when the camera angle changes, and it illustrates this with a 15-second virtual clothing try-on.

The company is also introducing conversational editing: you ask the model to change the background, lighting, weather, product or format of an entire clip, as if giving instructions to an editor. Google recommends making drafts at low resolution (360p or 720p) and upscaling only the final version to 1080p or 4K.

Not all of this showed up in external testing. The company Segmind made 20 API calls, costing US$16.17, and found that the editing endpoint, the technical gateway that activates that feature, returned "not available yet." It also reported that reference mode only reached 10 seconds with its access. Unite.ai, for its part, notes that the documentation uses 10 seconds by default. The feature exists on Google's blog, but it has yet to be confirmed that it is open to everyone.

On quality, the only comparative figures come from DeepMind itself: in its evaluations, 62% of raters preferred version 1.2 versus 38% who chose 1.1 for visual consistency. Google did not publish scores for lip sync, physics or reference fidelity. Segmind called it the most instruction-faithful model it had tested, although it missed roughly one in four instructions. It added that it does not consider it a model for people with cinematic realism, and that responses took between 28 and 136 seconds, usually between 40 and 90.

What a clip really costs

The 30% discount is measured against the previous version of the same model, not against the competition. Omni 1.1 would have cost around US$0.10 per second, a figure derived from the discount math and from a DEV Community article, but not verified on the official page.

According to Unite.ai, the API charges in tokens, the units the system uses to measure what it processes: US$12 per million tokens of generated video, at a rate of 5,792 tokens per second. That works out to about US$0.70 for a 10-second clip and US$1.04 for a 15-second one. The difference from the US$1.05 that results from the per-second rate is due to rounding. Input, which includes text, images and video, costs US$1.50 per million tokens, and there is no free tier.

A secondary blog compares prices at 720p: Veo 3.1 Fast at US$0.10 per second, Kling 3.0 Pro via Runway at US$0.17 with audio, and Sora 2 Pro at US$0.30. These figures have not been confirmed in official sources. Unite.ai adds that, starting December 3, calls to Omni 1.1 will automatically switch to version 1.2 at the new rate.

The videos carry SynthID, an invisible watermark for identifying generated content. One analysis warns that this watermark does not prove a scene is real, nor does it guarantee it will always be detected.

For a Latin American creator or small business, some questions remain open. Google has not published information on country-by-country availability in the region, billing in local currency or digital taxes. The price is in dollars and, with no free tier, an active paid account is required. In addition, the cost of upscaling the final version to 1080p or 4K is not included in what has been published, so the dollar per clip is only the starting point. It will be necessary to follow the official pricing table, independent testing, and how the model handles faces, brands and Spanish-language text before knowing whether it truly makes video production cheaper in the region.

Related news

Artificial Intelligence

Mathematician says he has proved the Riemann hypothesis with help from Claude; it still needs to be verified

Levent Alpöge has published a purported proof of one of the most famous problems in mathematics, produced together with Anthropic's AI. Early reviews are encouraging, but the community has not yet validated it.

Artificial Intelligence

OpenAI cuts GPT-6.2 Astra API pricing by 20%: what changes for those paying in dollars

OpenAI launched GPT-6.2 Astra and GPT-6.2 Sol. The flagship model's usage-based pricing drops, but the cheaper model's stays the same, and the fine print will determine how much users in Latin America actually save.

Artificial Intelligence

Nvidia reports earnings without Jensen Huang: what's at stake for AI chips

Five days after its founder's resignation, the company that dominates artificial intelligence chips faces Wall Street. For Latin America, open questions outweigh measurable effects for now.