Google Launches Nano Banana 2 Lite and Gemini Omni Flash API

Google has expanded its generative AI ecosystem with the release of two powerful new models designed to slash latency and unlock cinematic creativity. By introducing Nano Banana 2 Lite for high-speed image generation and Gemini Omni Flash for video synthesis, Google is providing developers with a streamlined pipeline to move from static concepts to moving media.

Nano Banana 2 Lite: High-Throughput Image Generation

Google is targeting the developer demand for speed and cost-efficiency with the launch of Nano Banana 2 Lite (technically identified in the API as gemini-3.1-flash-lite-image). This model is purpose-built for rapid ideation and high-volume production pipelines, capable of generating 1K resolution images in just four seconds.

At a price point of $0.034 per image, Nano Banana 2 Lite offers a significant economic advantage for companies running massive image-generation workflows. Despite its optimized speed, Google claims the model maintains high standards for prompt following, character consistency, and text readability. This model completes a three-tier hierarchy:

  • Nano Banana 2 Lite: Focused on extreme speed and low cost.
  • Nano Banana 2 (Gemini 3.1 Flash Image): The versatile "all-rounder."
  • Nano Banana Pro (Gemini 3.1 Pro Image): Designed for professional-grade reasoning and complex control.

The model is not just for developers; it is being integrated across the Google ecosystem, including Google Search (AI Mode), NotebookLM, Google Photos, and Google Ads.

Gemini Omni Flash: Bringing Video to the API

Following its preview at Google I/O, Gemini Omni Flash is now officially available via the Gemini API and Google AI Studio. This model represents a leap in multimodal reasoning, allowing developers to generate and edit video using natural language prompts. Priced at $0.10 per second of video output, it sits competitively alongside Veo 3.1 Fast.

Gemini Omni Flash excels at "conversational video editing," where users can refine clips using text, images, or existing video as input. While current limitations include a 10-second maximum clip length and limited character consistency across complex camera movements, the model’s ability to sync text and graphics with video actions marks a significant step forward for AI-driven content creation.

The Power of Model Chaining and the Interactions API

The true strategic value of this release lies in the synergy between the two models. Google is encouraging a "chaining" workflow: developers can use Nano Banana 2 Lite to instantly generate a high-quality reference image and then feed that image into Gemini Omni Flash to animate it into a video.

To facilitate this, Google is promoting its Interactions API, which serves as the default for these workflows. The API preserves session history and context, enabling up to three consecutive edits, making the creative process feel more like a continuous dialogue with the AI. To demonstrate these capabilities, Google has launched three demo apps: "Anywhere" for travel animation, "Space Lift" for interior design, and "Omni Product Studio" for e-commerce video generation.

As with all Google generative models, both Nano Banana 2 Lite and Gemini Omni Flash utilize SynthID watermarking to ensure AI-generated content can be verified across Google platforms.

Key Takeaways

  • Efficiency at Scale: Nano Banana 2 Lite delivers 1K images in four seconds for just $0.034, optimizing workflows for high-throughput needs.
  • Multimodal Video Editing: Gemini Omni Flash introduces natural language video generation and editing to the Gemini API at $0.10 per second.
  • Integrated Creative Pipeline: Developers can chain image and video models via the Interactions API to transform static assets into animated content seamlessly.