Google DeepMind Cuts Gemini Omni 1.2 Flash Pricing 30%, Extends Video Generation to 15 Seconds
Gemini Omni 1.2 Flash drops to $0.07/second for 720p video, adds native 15-second clips and improved consistency tracking across scenes.
Last verified:
Bottom Line Up Front
Google DeepMind released Gemini Omni 1.2 Flash on November 5, cutting pricing by 30% to $0.07 per second for 720p video while extending native generation from 10 to 15 seconds. The update addresses cost concerns through multi-turn editing workflows and stronger visual consistency, positioning the model for product-focused use cases like virtual try-ons and branded video content.
30% Price Reduction Targets Cost-Conscious Video Workflows
According to DeepMind, Gemini Omni 1.2 Flash now costs $0.07 per second for 720p output—a 30% decline from its predecessor. The pricing strategy mirrors a broader pattern in video-generation economics: vendors are lowering base rates while enabling cost reduction through architectural workflow changes rather than just hardware efficiency gains alone.
DeepMind’s design explicitly addresses generation inefficiency. Users can draft videos at 360p resolution first, then use multi-turn conversational editing to refine specific frames or sequences without re-generating the entire clip, and finally upscale completed versions to 1080p or 4K. This staged approach shifts computational cost to the user’s discretion—editing a single detail no longer requires re-running the full pipeline.
Extended Duration and Granular Timing Control
The model now supports native 15-second clips, up from the 10-second ceiling in Omni 1.1. DeepMind reports that duration can be specified in one-second increments from 3 to 15 seconds, allowing creators to produce complete product explainers, social-media ad spots, and narrative sequences in a single generation without concatenating multiple sub-clips.
This extension matters for downstream workflows in e-commerce and advertising, where 10-second windows often compress creative intent. A 15-second window accommodates camera pans, product rotations, and dialogue-driven storytelling more naturally.
Reference Adherence and Cross-Frame Consistency
The update emphasizes visual consistency as a primary advancement. DeepMind reports that Gemini Omni 1.2 Flash better maintains character identity, garment detail, product packaging, and scene lighting across camera transitions and motion sequences. The model can now combine reference images (static product photography, character sheets), short motion clips (up to three seconds), and audio cues while preserving fidelity from first to last frame.
The consistency improvements directly enable use cases that prior versions struggled with—virtual try-on sequences where a model’s face and body remain recognizable across 360-degree rotations, or branded catalog videos where product colors and typography survive camera pans and lighting shifts.
Why This Matters
The 30% price reduction, paired with iterative editing and extended duration, shifts the competitive calculus for small-to-medium production teams. Previously, cost per second incentivized longer clips and fewer iterations; now, the workflow economics favor rapid prototyping and refinement. Teams currently using Runway Gen-3 or Pika Labs for product video will face a re-evaluation trigger—particularly if the consistency improvements hold under external benchmarking.
The 15-second native window also raises the bar for competitors still capped at 10 seconds. However, the consistency claims are vendor-reported; independent evaluation against other models’ reference-following performance remains absent from public benchmarks. If consistency holds in production workflows, this feature may accelerate adoption among e-commerce and brand teams where visual fidelity directly affects conversion rates.
Frequently Asked Questions
What is the new pricing for Gemini Omni 1.2 Flash?
According to DeepMind, the price is $0.07 per second for 720p video generation, representing a 30% reduction from Gemini Omni 1.1.
How long can videos be generated with Gemini Omni 1.2 Flash?
The model supports native video creation up to 15 seconds, with duration specifiable in one-second increments from 3 to 15 seconds.
What new consistency improvements does Omni 1.2 Flash provide?
The model maintains stronger adherence to reference images (faces, clothing, product details, colors) and reduces visual drift across camera transitions and motion sequences.
What workflow changes does the new version enable?
Users can now draft at lower resolutions (360p), iteratively edit details without full re-generation, then upscale final versions to 1080p or 4K to control costs.