Google introduced the Nano Banana 2 Lite, an image model producing single 1024x1024 renders in roughly 4–5 seconds at $0,034.

Image generation and Comfy workflows

The model is exposed in Comfy with two dedicated workflows: one for text-to-image generation and another for editing images via textual prompts.

Output detail generally remains consistent at the declared resolution, but facial rendering can exhibit issues with resolution, lighting and compositional accuracy.

Video editing and Gemini Omni Flash

Comfy also added the Gemini Omni Flash video generator in an editing mode and an img-2-video workflow accepting reference frames and short clips.

The video tool accepts inputs up to 3 seconds and is priced at $0,10/sec, though generation may decline when people appear in the frame.

Availability across Google platforms

Besides Comfy, the models are available via Google AI Studio, the Gemini site and app, Flow studio, and to developers through the API.

Google Search’s AI Mode is reported to receive image-generation features, but that capability is not yet visible or globally deployed at this time.

Limitations and user experience

Users note that while the model maintains overall scene detail, it cannot produce resolutions above 1К, limiting use cases that require higher pixel density.

Practical constraints on faces and human subjects remain relevant for creative and production workflows, affecting final usability in some projects.