Google DeepMind has released Gemini Omni 1.2 Flash, an updated multimodal AI model that natively generates and edits video clips up to 15 seconds long. Available via Google Cloud and the Gemini API, the new model cuts video generation costs by 30% compared to version 1.1, pricing output at $0.07 per second for 720p resolution.

The model supports multi-turn conversational video editing through Google’s Interactions API, allowing developers and creators to alter background lighting, aspect ratios, or subjects without re-uploading or regenerating entire clips. The platform also integrates SynthID watermarking on output files and offers Provisioned Throughput options for enterprise customers.

Early enterprise partners such as invideo and QuickFrame AI reported improved prompt adherence and faster ad generation pipelines. Google Vids subscribers can access 15-second generation features at no additional charge.

Why it matters

  • Lowers economic barriers for commercial video generation down to $0.07 per second.

  • Enables longer 15-second continuous video generations natively without complex stitching.

  • Iterative API design preserves state, dramatically cutting redundant render costs for enterprise workflows.

Source: marktechpost.com