Google releases Gemini Omni 1.2 Flash video model: 15-second clips up to 4K with native audio

Google released Gemini Omni 1.2 Flash on November 5, 2026, a native multimodal video generation and editing model that succeeds Gemini Omni 1.1 Flash. According to its OpenRouter listing, it generates clips up to 15 seconds long with native audio, at resolutions from 360p drafts up to 4K.

Specifications

• Model: Gemini Omni 1.2 Flash (google/gemini-omni-1.2-flash)
• Released: November 5, 2026
• Context window: 131K
• Pricing: $1.50 per 1M input, $9 per 1M output
• Aspect ratios: 16:9 and 9:16
• Resolution: 360p drafts up to 4K
• Clip length: up to 15 seconds, with native audio

Capabilities

The listing describes support for:

• Text-to-video
• Image-to-video
• First and last frame interpolation
• Subject and video references, with what the listing calls improved reference management
• Editing or extending an uploaded video

Multi-turn editing and extension of a prior generation is supported through the previous_job_id parameter.

Access

Video generation is only supported through the Video Generation API, according to the listing. OpenRouter publishes a guide at openrouter.ai/docs/guides/overview/multimodal/video-generation.

Google AI Studio is the only provider listed, at $1.50 input and $9.00 output per 1M. OpenRouter reported a best-provider P50 end-to-end latency of 24.06 seconds when the page was read on November 5, 2026. The model is not currently available through EU in-region routing.

Predecessor

The OpenRouter listing for Gemini Omni 1.1 Flash describes it as the generally available successor to Gemini Omni Flash Preview. That listing cites support for up to three reference clips and up to 40 seconds of total video through multi-turn editing with previous_job_id. It shows the same $1.50 input and $9 output pricing and 131K context as the 1.2 listing. Benchmark scores for Gemini Omni 1.2 Flash are not published on the listing.

What this means

Gemini Omni 1.2 Flash continues the Gemini Omni line as a single model for both generating and editing video, with the 1.2 listing highlighting reference management as the changed area. For developers, the practical constraints are access and routing: the model runs through the Video Generation API rather than a standard text endpoint, Google AI Studio is the only listed provider, and EU in-region routing is not available. The listed per-1M pricing matches Gemini Omni 1.1 Flash, so teams migrating between versions can compare on the features listed rather than on the per-1M rate. The 15-second clip length and multi-turn previous_job_id workflow define how longer sequences would need to be assembled. The 1.1 listing cites a 40-second total limit, and the 1.2 listing does not state a total-length limit.