Google has announced two new generative media models aimed at developers: Nano Banana 2 Lite, a fast and cheap image generation model, and Gemini Omni Flash, which handles video generation and conversational video editing. Both are available now in Google AI Studio and the Gemini API.
The timing reflects a broader shift in how developers are building AI-powered products. Speed and cost have become just as important as raw quality, especially for apps that need to generate media at scale or in real time. Google is positioning these two models as the answer to that pressure.
The bigger picture here is that these models are designed to work together. Developers can generate an image with Nano Banana 2 Lite and then pass it directly to Gemini Omni Flash to animate it into a video, creating a full image-to-video pipeline without switching platforms.
Nano Banana 2 Lite: built for speed at scale
Nano Banana 2 Lite (model ID: gemini-3.1-flash-lite-image) is Google’s fastest image model to date. It generates text-to-image outputs in around 4 seconds and costs $0.034 per 1,000 images. For developers running high-volume pipelines or building interactive prototyping tools, those numbers matter.
Google is recommending that anyone still using the original Nano Banana model (gemini-2.5-flash-image) switch to Nano Banana 2 Lite immediately. The newer model is faster, cheaper, and produces better results.
Despite the speed focus, the model still holds up on the basics:
- Reliable prompt adherence
- Consistent character rendering across generations
- Legible text within images
It helps to understand where this model sits within the broader Nano Banana family:
- Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image): Fastest option, built for high-volume and near-real-time use cases
- Nano Banana 2 (Gemini 3.1 Flash Image): The balanced option, good quality at reasonable speed and cost
- Nano Banana Pro (Gemini 3 Pro Image): For complex work where accuracy matters more than speed
- Nano Banana (Gemini 2.5 Flash Image): The legacy model, now superseded
Beyond developer tools, Nano Banana 2 Lite is also rolling out inside Google’s consumer products, including AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow, and Google Ads.
Gemini Omni Flash: video generation meets conversational editing
Gemini Omni Flash (gemini-omni-flash-preview) was first shown at Google I/O and is now available to developers via the Gemini API and Google AI Studio. It supports video generation and editing from text, image, and video inputs combined.
Pricing is set at $0.10 per second of video output, which matches what Google charges for Veo 3.1 Fast.
The model’s main strengths include:
- Conversational video editing: You can refine a video by describing changes in plain language, no timeline scrubbing required
- Multimodal referencing: Feed it a mix of images, text, and video clips to control what the output looks like
- Real-world knowledge: The model draws on Gemini’s broader knowledge base, including history, biology, and narrative structure, to produce videos that make sense
- Text and action sync: Graphics and on-screen text can be tied to specific actions in the video through simple prompts
There are some real limitations worth knowing about before you build:
- Generations are capped at 10 seconds for now, with longer durations coming later
- Audio references and scene extension are not yet supported via the API
- Video references up to 3 seconds are accepted by the schema but are not correctly processed at this time
- Character consistency across scene changes and panning shots is still being worked on
Chaining the two models together
Google has put together three demo apps to show what a combined Nano Banana 2 Lite and Gemini Omni Flash workflow looks like in practice.
Anywhere lets you upload a selfie and uses Nano Banana 2 Lite to place you at famous landmarks around the world. Click any generated image and Omni Flash turns it into an animated video clip of that location.
Space Lift is an interior design app. Upload a photo of a room and Nano Banana 2 Lite generates redesigns across different styles. Pick one you like and Omni Flash produces a short cinematic video walkthrough of the new design.
Omni Product Studio converts static product images created by Nano Banana 2 Lite into short e-commerce videos using Gemini Omni Flash, which is a useful workflow for anyone building retail or marketing tools.
Developers using the Interactions API with these models can maintain session history across multi-turn conversations, allowing users to stack up to three sequential edits while keeping context intact.
Safety and content verification
Both models use SynthID watermarking, Google’s system for embedding invisible markers in AI-generated content. Users can verify whether content was AI-generated through the Gemini app, Gemini in Chrome, or Search. Google says it is expanding these verification tools across the web to help people understand how content was created or edited.
