Google has released Gemini Omni 1.1 Flash (gemini-omni-1.1-flash-preview), bringing expanded temporal context windows, reference conditioning, and tiered generation pricing to its multimodal video generation API.
The model is accessible immediately through Google AI Studio and the Gemini Enterprise Agent Platform, supporting developers targeting programmatic video synthesis, interactive media pipelines, and dynamic storyboarding.
Extended Temporal Conditioning and Keyframe Controls
The primary architectural update in version 1.1 focuses on continuity across sequence boundaries:
- 10-Second Context Buffering: Rather than conditioning subsequent frames on single-frame endpoints, the model ingests up to 10 seconds of prior video history when computing temporal extensions.
- Incremental Scene Chaining: Sequences can be extended in 10-second blocks up to a total continuous duration of 40 seconds while preserving lighting consistency, entity geometry, and motion dynamics.
- First and Last Frame Interpolation: Developers can supply explicit starting and ending keyframes, prompting the model to generate smooth visual and camera transitions between two defined states.
- Reference Video Conditioning: Up to 3 seconds of reference video can be injected into the prompt context to anchor character appearance, stylistic grading, or specific motion motifs across novel generations.
Multi-Tier Resolution and Inference Economics
To reduce iteration latency during prototyping, Google introduced a low-resolution draft mode alongside native high-resolution rendering and upscaling paths.
| Resolution Tier | Latency Profile | Cost per Second | Use Case Target | | :--- | :--- | :--- | :--- | | 360p Draft | Up to 60% faster generation | $0.03 / sec | Rapid prototyping, prompt tuning, layout validation | | 720p Standard | Standard baseline | $0.10 / sec | Real-time web previews, interactive agent feeds | | 1080p HD | High-fidelity rendering | $0.15 / sec | Production digital delivery, social distribution | | 4K Ultra HD | Upscaled neural reconstruction | $0.30 / sec | Broadcast mastering, high-resolution visual assets |
Context Limits and API Specifications
The gemini-omni-1.1-flash-preview endpoint operates with a 131,072-token input window and generates up to 57,920 output tokens. Supported container formats include MP4, WebM, QuickTime, and standard streaming containers across standard 16:9 landscape and 9:16 vertical aspect ratios.
Integrations are already live across third-party creative tooling platforms, including Figma Weave, with rollout planned for consumer-facing Google AI subscription tiers.
