Nano Banana 2 Lite becomes the newest addition to Google’s Gemini Image family and is positioned as its fastest and most cost-efficient image model. Google recommends it as the replacement for the original Nano Banana (gemini-2.5-flash-image), saying it offers improvements in speed, quality, and cost.
The company says Nano Banana 2 Lite generates text-to-image outputs in approximately four seconds and is priced at $0.034 per 1,000-resolution image. While prioritizing low latency and affordability, Google says the model maintains reliable prompt adherence, character consistency, and legible in-image text rendering, making it suitable for rapid prototyping and high-volume developer workloads.
Google also outlined how the Nano Banana model family is now structured. Nano Banana 2 Lite is optimized for high-speed workflows, Nano Banana 2 is designed as the general-purpose option balancing quality and performance, Nano Banana Pro targets professional use cases requiring greater reasoning and control, and the original Nano Banana remains available as a legacy model.
Alongside the image model, Google is opening Gemini Omni Flash to developers after first introducing it at Google I/O. The multimodal model supports video generation and editing using combinations of text, images, and video as inputs.
Google says Gemini Omni Flash is priced at $0.10 per second of generated video, matching the pricing of Veo 3.1 Fast. The model supports conversational video editing through natural language, multimodal referencing that combines different media types to maintain scene consistency, synchronization between text and on-screen actions, and video generation that draws on Gemini’s broader knowledge.
The current preview includes several limitations. Video generation is limited to 10-second clips, with longer durations planned for a future release. Audio reference uploads and scene extension are not yet supported through the Gemini API, while video references up to three seconds are accepted but are not currently processed correctly. Google also says character consistency during scene changes and camera movements remains an area under active development.
To support more complex creative workflows, Google is highlighting the Interactions API, which preserves session history across multiple editing steps and allows developers to chain together up to three sequential edits. The company says this enables developers to move from image generation to video creation within a single workflow while maintaining conversational context.
Google is also releasing several demonstration applications built with the new models. Anywhere places users into landmark locations using Nano Banana 2 Lite before animating the results with Gemini Omni Flash. Space Lift reimagines interior spaces and produces cinematic walkthroughs, while Omni Product Studio converts static product images into e-commerce videos.
“We’re making it easier to experiment and scale your ideas with Nano Banana 2 Lite, our fastest, most cost-efficient Gemini Image model, and Gemini Omni Flash for high-quality video generation and conversational editing,” said Alisa Fortin, Product Manager, Google DeepMind.
Both models include SynthID watermarking, Google’s AI content provenance technology. The company says generated content can be verified through the Gemini app, Gemini in Chrome, and Google Search as part of its broader effort to provide transparency around AI-generated media.
This analysis is based on reporting from Google.
Image courtesy of Google.
This article was generated with AI assistance and reviewed for accuracy and quality.
