Google releases Nano Banana 2 Lite image model and Gemini Omni Flash video model

Google DeepMind unveiled two new Gemini‑family generative AI models. Nano Banana 2 Lite (gemini‑3.1‑flash‑lite‑image) creates 1 K‑resolution images in about four seconds and costs roughly $0.034 per 1 000 images, a five‑fold speed boost and half the price of the prior Nano Banana version. Gemini Omni Flash (gemini‑omni‑flash‑preview) generates short video clips (up to 10 seconds) at about $0.10 per second, supporting text, image and audio inputs and offering basic editing via prompts.

Both models are available through Google AI Studio, the Gemini API and the Gemini Enterprise Agent platform, and embed SynthID watermarks for provenance tracking. Google extended Gemini’s personalized image generation – powered by Nano Banana – to free users in the United States, letting the service draw on Personal Intelligence data from Gmail, Photos, YouTube and Search when users grant permission.

The rollout also includes several downstream features: the Gemini Spark agentic assistant now runs on macOS, integrates with Google Workspace and third‑party apps, and can act on remote commands; Ask Photos conversational photo editing is being expanded to Android users in Germany, the United Kingdom, France, Spain and Italy; NotebookLM can produce 60‑second, vertical “short video overviews” using Nano Banana 2 Lite; and Gemini in Google Slides enables single‑prompt creation of multi‑slide decks with contextual grounding from Drive files. Limitations such as the 10‑second video cap, language‑only support for some features, and age or subscription requirements remain in place.