Nano Banana 2 Lite and Gemini Omni Flash: Google's New AI Models for Images and Video

On June 30, 2026, Google released two new generative AI models: Nano Banana 2 Lite, the fastest and most affordable image model in the Nano Banana series, and Gemini Omni Flash, a model for video generation and video editing via voice commands. Both are now available via Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform. Nano Banana 2 Lite generates an image in about four seconds and costs $0.034 per 1K image, while Omni Flash costs $0.10 per second of generated video.

Key Takeaways

Nano Banana 2 Lite (technically: Gemini 3.1 Flash Lite Image) is designed for speed and volume and is intended to replace the original Nano Banana model.
Gemini Omni Flash is now available to developers for the first time and creates and edits videos from text, image, and video inputs.
Pricing: $0.034 per 1K image (Nano Banana 2 Lite), $0.10 per video second (Omni Flash, same price as Veo 3.1 Fast).
Available in Google AI Studio, the Gemini API, and the Gemini app; Nano Banana 2 Lite is also available in Search’s AI Mode, NotebookLM, Google Photos, and Google Ads.
All outputs bear an invisible SynthID watermark.

What is Nano Banana 2 Lite?

Nano Banana 2 Lite is Google’s new entry-level model for AI image generation with Gemini. It’s aimed at anyone who needs a lot of images quickly while keeping costs in mind. The official model name is Gemini 3.1 Flash Lite Image; “Nano Banana” is the catchy codename that Google now openly uses for its native image features.

Two features stand out. First, the speed: An image is generated from a text prompt in about four seconds. This makes the model practical for quick experimentation, for drafts, and for applications where users are watching in real time. Second, the price: $0.034 per image at 1K resolution. Anyone generating images on a large scale will quickly notice this difference in their bill.

Image: Google

Despite the focus on speed, Google promises that the basics are solid. The model is designed to reliably implement prompts, maintain character consistency across multiple images, and ensure that text in the image is legible. It is precisely this last point where many image AI systems have long fallen short. According to Google, anyone still using the first Nano-Banana model (Gemini 2.5 Flash Image) can switch over immediately and enjoy better quality, faster performance, and lower costs.

The Nano-Banana Family at a Glance

With the new Lite model, the Nano-Banana series now consists of four variants. Which one you choose depends on whether speed, balance, or maximum control is more important to you.

Model	Technical Name	Designed for
Nano Banana 2 Lite	Gemini 3.1 Flash Lite Image	Speed and capacity, very low latency
Nano Banana 2	Gemini 3.1 Flash Image	All-rounder with a good balance of quality, speed, and price
Nano Banana Pro	Gemini 3 Pro Image	Complex, professional tasks requiring a high degree of control and reasoning
Nano Banana (Legacy)	Gemini 2.5 Flash Image	First model in the series; upgrade recommended

The logic behind this is simple: Lite for quick drafts and high volumes, the standard model for everyday use, and Pro when accuracy is more important than speed. In addition to developer access, Nano Banana 2 Lite is also rolling out in Google’s consumer products, including AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow, and Google Ads.

Gemini Omni Flash: Create and Edit Videos Using Natural Language

The second new addition is the more exciting one. Gemini Omni Flash is a model for video generation and—here’s the real kicker—for conversational video editing. This means you can use natural language to describe what should change in a clip, rather than editing on a timeline. Omni was already unveiled at Google I/O 2026, where Google positioned the Gemini app as a much more agent-like tool overall. Now, the Flash model is available to developers for the first time via the Gemini API and Google AI Studio.

Omni Flash accepts a combination of text, images, and video as input. Google highlights four key capabilities in particular:

Conversational editing: Customize videos using natural language, without a traditional editing program.
Multimodal references: Combine images, text, and video to maintain consistency in a scene.
World knowledge: The model draws on Gemini’s knowledge of topics such as history, biology, and narrative logic.
Synchronization: Text and graphics can be directly linked to movements in the video via a prompt.

In terms of pricing, Omni Flash costs $0.10 per second of generated video—according to Google, the same as Veo 3.1 Fast. However, the launch isn’t without its caveats. Currently, the model only generates clips that are ten seconds long; longer videos are expected to follow. The API does not yet support audio references or extending scenes. Although the interface accepts video references up to three seconds long, it does not currently process them correctly. And during scene changes or camera pans, the model does not always maintain consistent character rendering. While this is understandable for a public preview, you should still plan accordingly.

Combining Both Models

Things get interesting when you chain the two models together. Here’s the workflow Google suggests: First, use Nano Banana 2 Lite to generate an image in a flash; then pass it as a reference to Gemini Omni Flash and have it animated into a video. The Interactions API preserves the context of a session, allowing users to build upon up to three edits in succession.

To try this out, Google provides three demo apps in AI Studio that showcase this interaction: “Anywhere” places an uploaded photo in well-known locations and animates the result; “Space Lift” redesigns rooms and displays them as a short clip; and “Omni Product Studio” turns static product images into short e-commerce videos. For anyone building their own AI applications, this is a handy template.

What does this mean for you in Germany?

The short answer: Yes, you can use it—with a few limitations. Nano Banana 2 Lite and Omni Flash run via Google AI Studio, the Gemini API, and the Gemini app, and these options are available here as well. Pure image generation with Nano Banana is available in all countries where the Gemini app is available.

The situation is different when it comes to highly personalized image creation. This feature links Nano Banana to Google’s Personal Intelligence and accesses Google Photos, Gmail, and your search history to do so. This feature is currently limited to the U.S. and is not available in Europe, Switzerland, or the U.K., presumably for data protection reasons.

For transparency: Every image or video generated with the new models receives an invisible SynthID watermark. You can check whether content comes from Google’s AI via the Gemini app, Gemini in Chrome, or Search. It’s not perfect, because SynthID only works with Google’s own models, but it’s at least a step toward combating the growing flood of AI-generated media.

This dual launch can also be viewed in the context of the broader AI race. While Google is expanding its image and video capabilities, chatbot providers are simultaneously locked in their own battle—see, for example , Anthropic’s Claude Sonnet 5, which is closing in on the top model in terms of price. The pace at which new models are being released has noticeably increased in 2026.

Frequently Asked Questions About Nano Banana 2 Lite and Gemini Omni Flash

What is Nano Banana 2 Lite?
Nano Banana 2 Lite is Google’s fastest and most affordable AI image model from the Nano Banana family. It generates images in about four seconds and costs $0.034 per 1K image. Its technical name is Gemini 3.1 Flash Lite Image.

What can Gemini Omni Flash do?
Gemini Omni Flash creates and edits videos from text, image, and video inputs. Editing is controlled via natural language. Currently, the model generates clips up to ten seconds long.

How much do the new models cost?
Nano Banana 2 Lite costs $0.034 per 1K image. Gemini Omni Flash costs $0.10 per second of video generated, the same as Veo 3.1 Fast.

Are Nano Banana 2 Lite and Omni Flash available in Germany?
Yes. Both models run on Google AI Studio, the Gemini API, and the Gemini app, all of which are accessible in Germany. Only personalized image generation via Personal Intelligence remains limited to the U.S. for now.

How can I tell if an image isgeneratedby Google’s AI?
All content generated with these models bears an invisible SynthID watermark. You can check this using the Gemini app, Gemini in Chrome, or Google Search.

Conclusion

With Nano Banana 2 Lite and Gemini Omni Flash, Google is consistently focusing on fast, affordable building blocks for creative AI applications. The Lite image model is particularly exciting for developers and heavy users, as a processing time of four seconds per image and a cost of three cents make new applications economically viable. Omni Flash is the real showstopper, but as a preview, it has visible limitations: ten-second clips, a few missing API features, and inconsistencies. If you want to start experimenting today, the demo apps are a great place to begin. And those who prefer to wait will likely see the next update in the coming months anyway, given how quickly the Nano Banana series is currently evolving.

Sources and further information: Google Keyword Blog, Google DeepMind on Gemini Omni, and the Gemini API documentation.