Google Deepmind extends Gemini Omni 1.2 Flash video duration

Google Deepmind is releasing its Gemini Omni 1.2 Flash video model for general use, according to a report by The-decoder. This update increases the maximum length of generated video clips from 10 seconds to 15 seconds. Matthias Bastian reports that the cost for 720p video has decreased by 30 percent to $0.07 per second. The new model also introduces a conversational editing feature to allow for iterative changes to clips. Google is rolling this out across its API, Cloud Console, and various creative applications.

Why it matters

Lowered costs and longer clip durations change the economics of AI-driven video production. For businesses, the ability to refine specific details through conversation rather than starting from scratch reduces wasted compute time. This allows for a more controlled production pipeline where visual consistency is a requirement rather than a variable. The integration of reference images means brand assets can be used more reliably.

• Reduced overhead for creating short-form social media content.
• Higher precision in maintaining brand colours and typography across clips.
• Faster prototyping of video concepts through low-resolution drafts.

What to watch

You should test how the model handles specific brand assets like packaging typography and exact colour hex codes. While Google claims better adherence to reference images, the practical utility for high-end video production depends on the precision of these outputs. We recommend auditing the visual drift during complex camera movements before integrating these tools into a professional content workflow.

Within a year, video editing will shift from frame manipulation to natural language instruction. This requires model architectures to master temporal consistency across long sequences.