Google Releases Gemini Omni 1.2 Flash: 15-Second Videos, Conversational Editing, and 30% Price Cut

Google launched Gemini Omni 1.2 Flash on November 5, the latest version of its video generation and editing model. Three changes stand out for creators: clips expanding from 10 to 15 seconds, conversational editing on entire clips, and an API price cut of 30%, down to $0.07 per second in 720p. For the first time, an Omni model has reached general availability, meaning it is ready for production use backed by Google Cloud enterprise guarantees.

15 Seconds, Adjustable to the Exact Second

Omni 1.1 Flash capped generations at 10 seconds. Version 1.2 extends that to 15 seconds, with a selectable duration between 3 and 15 seconds in one-second increments. Google positions this as the ideal length for a social media ad or a short explainer video in a single generation pass. API release notes detail other boundaries: interpolation between a first and last frame spans 3 to 10 seconds, and clip extensions run in increments of 3 to 10 seconds, up to 30 seconds total.

Editing Instead of Regenerating from Scratch

Multi-turn editing now works across full 15-second clips: users can swap a product, alter scenery, adjust lighting, change the weather, or modify the aspect ratio through sequential prompts without starting over. The model better adheres to provided references: character, product, or style images (up to six images according to the documentation), short motion snippets of up to three seconds, and audio cues. Google promises faces, clothing, packaging, and brand colors that stay consistent across camera angles, more natural dialogue with accurately attributed voices, and fewer jarring artifacts during continuous camera moves.

Pricing, Throughput, and Migration

Via the API, pricing drops to $0.07 per second in 720p, or $12 per million video output tokens, while paid rate limits double. Google recommends drafting at 360p or 720p, then upscaling the best shots to 1080p or 4K. A key note for developers: starting December 3, 2026, calls to gemini-omni-1.1-flash will automatically redirect to version 1.2 at version 1.2 rates.

On the product side, the model is rolling out to Flow for directing and editing, to Google Vids (15-second clips at no extra cost), and to Google Ads; in the Gemini app, 15-second videos are currently rolling out to Google AI Ultra subscribers. All generated videos carry the invisible SynthID watermark.

What We Don’t Know

• Real-world quality: Examples are curated by Google, with no independent comparisons published yet.
• Limitations: Google itself acknowledges that progress is still needed on fast-paced sports action, rapid script changes, and complex physics.
• Wider access: There is no timeline for expanding 15-second generations in the Gemini app beyond Ultra subscribers.