Google has launched two new AI models built for speed and affordability: Nano Banana 2 Lite for image generation and Gemini Omni Flash for video creation. These models represent a shift toward practical, cost-effective AI tools that prioritize quick turnaround times over maximum quality, making them ideal for rapid prototyping and content creation at scale.

What Are These New Google AI Models?

Nano Banana 2 Lite, officially named Gemini 3.1 Flash-Lite Image, is an image generation model based on Google's Gemini 3.1 Flash-Lite architecture. It can process up to 1 million tokens of context and generate images from text descriptions in approximately 4 seconds, compared to 20 seconds for its predecessor, Nano Banana 2. The model outputs both images and text, with a maximum of 4,000 tokens for image output and 64,000 tokens for text output.

Gemini Omni Flash, currently in public preview, handles video generation and editing by combining text, images, and videos as input. It supports interactive editing using natural language instructions, allowing users to correct videos with written commands and maintain scene consistency by referencing images and text. The model can also synchronize on-screen text and graphics with the movements of people and objects in videos.

How Do Speed and Cost Compare to Existing Models?

The pricing and performance metrics reveal Google's strategy of balancing quality with accessibility. Nano Banana 2 Lite generates a 1K resolution image for approximately 3 cents, compared to 3.9 cents for Nano Banana 2 and 6.7 cents for the original Nano Banana model. On quality benchmarks, Nano Banana 2 Lite scored 1,251 on image generation tests, falling 19 points short of Nano Banana 2's 1,270 but exceeding the original Nano Banana's 1,151 score.

Gemini Omni Flash costs 10 cents per second of 720p video output, placing it on par with Google's speed-focused video model, Veo 3.1 Fast. Currently, the model generates videos up to 10 seconds long, though Google plans to support longer outputs in the future. For developers, the API pricing for Nano Banana 2 Lite is 25 cents per 1 million input tokens and $1.50 per 1 million output tokens.

Where Are These Models Available?

Both models are accessible through multiple Google platforms. Nano Banana 2 Lite is available in Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform, with rollout planned for consumer services including Google Search's AI Mode, the Gemini app, NotebookLM, Google Photos, Google Ads, and other products. Gemini Omni Flash is available to developers through Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform, with support for Gemini apps and Google Flow.

The most immediate consumer application is NotebookLM, Google's research tool, which now features "Short Video Overviews." These are 60-second portrait-oriented videos that automatically summarize a single concept from your uploaded documents, PDFs, or notes. The rollout began on June 30, 2026, initially for Google AI Ultra and Pro subscribers on mobile and web, with free users gaining access at a later date.

How to Use These Models for Content Creation

• Image Generation Workflow: Upload a text prompt to Google AI Studio or the Gemini API using Nano Banana 2 Lite to generate images in approximately 4 seconds. The model supports consistent character generation across multiple images and produces legible text within images, making it suitable for design reviews and rapid prototyping.
• Video Creation Process: Use Gemini Omni Flash to combine text, images, and video references to generate new video content with native audio. You can edit videos using natural language instructions, such as correcting scenes or adjusting visual elements while maintaining consistency.
• Educational Content Production: In NotebookLM, upload a source document and use the Short Video Overviews feature to automatically generate a 60-second vertical video explaining a concept. The system pairs explanatory animations with an AI narrator, creating mobile-friendly learning content.

What Are the Current Limitations?

Google openly acknowledges several constraints with Gemini Omni Flash. The model currently generates only 10-second clips, and the API does not yet support uploading audio references or extending scenes. While the API accepts video references up to 3 seconds long, the model may not process them correctly. Character consistency can break during scene changes and camera movements, meaning the model is not yet intended for production use without careful review of each generated clip.

For NotebookLM's Short Video Overviews, the feature launches only in English, though Google expects additional languages to follow within weeks. The feature is initially rolling out to paid subscribers, with a timeline for free user access not yet specified.

Why Does This Matter for Developers and Content Creators?

These models represent Google's commitment to democratizing AI-powered content creation by prioritizing speed and cost over maximum quality. Nano Banana 2 Lite achieves a balance between quality, speed, and price, making it practical for high-volume processing where rapid iteration matters more than perfection. This approach enables developers to build applications that generate images and videos at scale without prohibitive costs.

For NotebookLM users, the Short Video Overviews feature transforms how research and learning content is consumed. Google is positioning the tool with the slogan "Doom scrolling, but make it educational," acknowledging that the format mirrors social media consumption habits while delivering substantive learning content. The integration of Nano Banana 2 Lite and Gemini Omni Flash into NotebookLM demonstrates how Google is interlinking its AI services to create seamless workflows from document upload to polished video output.

All images generated by Nano Banana 2 Lite are embedded with an invisible digital watermark called SynthID that identifies them as AI-generated, reflecting Google's commitment to transparency as the volume of generated images increases.