Google Launches Two Models on the Same Day: Gemini Omni Flash and Nano Banana 2 Lite

On June 30, 2026, Google announced two new generative models simultaneously: Gemini Omni Flash and Nano Banana 2 Lite. Both are immediately available to developers through Google AI Studio and the Gemini API, covering different modalities — video for the first, images for the second — with a clearly defined pricing structure from day one.

Google Announces a Video and Image Model in a Single Day

Simultaneously releasing two models covering different modalities is a rare move even for Google. Gemini Omni Flash is aimed at video producers and developers who need fast, conversational editing of clips using natural language. Nano Banana 2 Lite targets mass image production where speed and low costs are the key factors. The common denominator: both models target high-volume, cost-sensitive use cases in which predictable pricing is as important as output quality.

What Does Gemini Omni Flash Offer?

Gemini Omni Flash (gemini-omni-flash-preview) is a multimodal model that combines Gemini’s reasoning capacity with video generation and editing. It accepts inputs in the form of text, images, and video, and enables conversational clip refinement using natural language — the user describes a change, and the model applies it without regenerating from scratch.

Clips are capped at a maximum of 10 seconds. The price is $0.10 per second of video output, meaning a ten-second clip — the maximum length — costs exactly $1.00. Longer durations have been announced but without a confirmed date.

The model is available in Google AI Studio, the Gemini API, the Gemini Enterprise Agent Platform, the Gemini app, and Google Flow. Use cases include automated short-ad production, Stories format, promo clips, and rapid prototyping of creative concepts.

Google explicitly states the limitations at this stage. Audio references are not yet supported. Scene extension is not possible via the API. Video references longer than 3 seconds are not processed consistently. Character consistency when changing scenes or during panning camera movements can be problematic — particularly for characters that appear across multiple clips within the same production.

Nano Banana 2 Lite: Faster, Cheaper, More Widely Available

Nano Banana 2 Lite (gemini-3.1-flash-lite-image) is the only one of the two models with GA status — generally available, with no preview restrictions. Google positions it as the fastest and most cost-effective Gemini Image model in its current lineup.

The key figure: text-to-image latency is just 4 seconds. The price is $0.034 per 1,000 images at 1K resolution. For context, that is under 3.5 cents for a thousand generated images — placing it among the most affordable commercial options for mass visual content synthesis.

The model supports image editing, generation of readable text within images, and character consistency across multiple successive generations in the same session. Distribution is broader than Omni Flash: in addition to Google AI Studio and the Gemini API, Nano Banana 2 Lite is integrated into AI Mode in Google Search, NotebookLM, Google Photos, and Google Ads — meaning many users are already interacting with the model without explicitly knowing the implementation.

Market Context and Limitations

Google’s move with Nano Banana 2 Lite is a direct response to Stability AI and Ideogram in the segment of cheap image models for high-volume use cases. A latency of 4 seconds and pricing of $0.034 per thousand images set a new reference point within the Gemini platform for teams working with large volumes of generated imagery.

Gemini Omni Flash is entering a space where OpenAI has not yet deployed a stable video solution in a production API. The conversational approach to editing — the user describes what they want changed, the model applies the change — alters the production workflow for small studios and creators who previously could not afford the time or financial cost of traditional video tools.

The 10-second limit excludes Omni Flash from the longer video content category covered by models such as Veo 3 and Sora 2, but for short formats — ads, story clips, teaser content — this is a natural fit with a clear and transparent pricing model. Longer durations, audio references, and improved character consistency across scene changes have been announced for future iterations.