Gemini Omni 1.2 Flash lowered its video price

Google DeepMind has launched Gemini Omni 1.2 Flash, its video generation model. The model has three standout upgrades: up to 15 seconds of video in a single pass, scene editing through chat, and a 30 percent lower price than the previous version.

7 cents per second

Video output at 720p resolution is priced at $0.07 per second. That brings the video token cost down to around $12 per million tokens. Google says this is a 30 percent discount compared to Omni 1.1 Flash.

The price cut becomes especially meaningful when paired with the draft-to-upscale workflow. Developers can first test a scene at 360p or 720p, and once the camera movement and composition look right, upscale only the final frames to 1080p or 4K. That significantly reduces the cost of trial and error.

Chat-based editing preserves the scene

The model’s most striking feature is multi-turn editing. In older models, even a small requested change in a generated video meant the clip had to be regenerated from scratch, and often the character or location would change too. With its multimodal memory layer, Omni 1.2 Flash preserves the scene’s geometry and applies only the requested change. For example, commands like “make the jacket leather, but don’t change the walking pace” or “slowly tilt the camera upward” work on the same shot.

Length increased to 15 seconds, audio is embedded in the video

The clip limit has been raised from 10 seconds to 15 seconds. Length can be selected in one-second increments between 3 and 15 seconds, which makes it a natural fit for vertical formats like TikTok, Reels, and Shorts.

On the audio side, there’s no need for a separate library: wind, footsteps, ambient hum, and multi-speaker dialogue are embedded directly into the video in sync with the action in the scene (native audio).

Where can you use it?

The model is available through Vertex AI, Google AI Studio, and the Gemini API. With this, Google is taking on OpenAI Sora and Runway in the video generation market with both pricing and editing capabilities.

Source: WebTuring