Google Unveils “Nano Banana 2 Lite” and “Gemini Omni Flash” for Multimedia Development

Nano Banana 2 Lite is a cost-effective image model optimized for developers who prioritize throughput and speed. It replaces the first-generation Nano Banana model and offers significantly lower latency, delivering text-to-image results within four seconds. The model costs $0.034 per 1K-resolution image, making it suitable for rapid prototyping and large-scale design work. Despite its focus on speed, the model retains the ability to render legible text and ensure character consistency across all images. This model is part of a larger model family that also includes the balanced “Nano Banana 2” and the high-precision “Nano Banana Pro.”

The image model is complemented by “Gemini Omni Flash,” a multimodal tool capable of video generation and interactive editing. It enables developers to refine video content using natural language inputs and combine multiple input types—including text and static images—to maintain control over the scene. The price for the model is $0.10 per second of output. Although the model is powerful, there are currently technical limitations, such as a maximum clip length of 10 seconds and certain inconsistencies in character consistency during complex camera movements like pans.

A key aspect of these releases is the ability to chain the two models together. Developers can use “Nano Banana 2 Lite” to quickly create a visual concept and then pass that image on to “Gemini Omni Flash” as a template for the animation. To support this iterative process, the Interactions API allows users to maintain session history and context across up to three consecutive editing steps.

To address concerns about AI-generated content, Google has integrated the SynthID watermark into both models. This ensures that the results are recognizable as AI-generated; corresponding verification tools are available via Search, Chrome, and the Gemini app.

Both Nano Banana 2 Lite and Gemini Omni Flash are currently available to developers via Google AI Studio and the Gemini API, while the image model is also being integrated into consumer products such as NotebookLM and Google Photos.