Gemini Omni 1.2 Flash Released with Lower Pricing and SLA

Google DeepMind has released Gemini Omni 1.2 Flash, a new version of its video generation and editing model. The most significant aspect of this release is not just visual quality, but the fact that the Omni family is receiving general availability status with official SLA guarantees for the first time.

• A one-second 720p video now costs $0.07, which is 30% cheaper than the previous version.
• Maximum video duration has been increased from 10 to 15 seconds.
• A full minute of finished 720p video now costs approximately $4.20.

New Features and Capabilities

Omni 1.1 Flash was released in late August, priced around $0.10 per second of 720p video. Version 1.2 phases out the preview stage with lower costs, making it significantly more affordable for marketing teams and developers creating multiple creative options. Starting December 3, requests to version 1.1 will automatically route to 1.2 at the new pricing, meaning developers won't need to rewrite their integration pipelines.

The model accepts prompts, character and product reference images, short video clips of up to three seconds, and audio cues. Google promises better object consistency from the first frame to the last, improved lip-syncing with multiple speakers, and fewer random cuts during complex camera movements like FPV shots or orbital tracking. Furthermore, users can edit videos conversationally by changing the background, lighting, weather, or aspect ratio without regenerating from scratch.

Pricing, Limitations, and Industry Impact

API billing is token-based: output costs $12 per million tokens with 5,792 tokens required per second of 720p video, while input costs $1.50 per million tokens. There is no free tier available. Limitations include difficulties with fast sports scenes, sudden scenario changes, and complex physics, while the API officially supports English only.

Internal Google DeepMind evaluations show that version 1.2 outperforms 1.1 in 62% of cases for visual consistency and 63% for identity preservation across multiple references. Partners such as Runway, Krea, Picsart, and QuickFrame are integrating the technology to reach massive audiences. This shift toward lower pricing and infrastructure-grade SLA indicates Google's strategic focus on high-volume enterprise and developer markets where cost efficiency matters most.