Google Gemini Omni 1.2 Flash Guide

Gemini Omni 1.2 Flash is Google DeepMind’s video creation and editing model, released and generally available on November 5, 2026. It makes clips up to 15 seconds long with synchronised audio and lets you edit them by talking to it.

Generate a clip, then keep editing it in conversation instead of starting over.

Length

Up to 15 seconds, up from 10; pick any length from 3 to 15 seconds

Editing

Multi-turn edits: swap products or backgrounds, change lighting or weather, or switch aspect ratio

Price

30% lower, at $0.07 per second for 720p in the API

References

Keeps characters, products, and styles from your references more faithfully

Speech

More natural dialogue, clearer lip-sync, and multiple speakers

Stability

Less visual drift and fewer unwanted jump cuts

• Text prompts
• Reference images of characters or products
• Short motion clips, up to three seconds
• Audio cues
• Your earlier clips, for multi-turn edits

Gemini app

15-second videos, rolling out to Google AI Ultra subscribers

Flow

Directing scenes and multi-turn edits

Google Vids

15-second clips at no extra cost with a Google account or Workspace

Google Ads

Scaling video ad campaigns

Gemini API and Vertex AI

For developers and businesses

• Draft at 360p or 720p to save money
• Refine with conversational edits
• Upscale the best result to 1080p or 4K
• Export the aspect ratio you need

A 12-second product video: a matte black water bottle on a wet rock by a mountain stream at sunrise, slow push-in, mist rising, gentle water sounds, ending on the bottle’s logo in sharp focus.

• Describe subject, setting, action, camera, and sound
• Use reference images for anything that must stay exact
• Make one change per edit turn

• “Make it rain and add thunder.”
• “Swap the bottle for the green version.”
• “Change to vertical 9:16.”
• “Make the lighting warmer.”

• Product explainers and catalogue spins
• Virtual try-on clips
• Branded social ads and ad variants for different audiences
• Short training and onboarding clips

• Google says fast sports action, rapid script changes, and complex physics are still being improved
• Clips longer than 15 seconds need stitching
• Text and logos still need checking in every frame

• Every output carries Google’s invisible SynthID watermark
• Label AI video where platforms or laws require it
• Use only likenesses, products, and audio you have rights to