Start building with Nano Banana 2 Lite and Gemini Omni Flash

Google introduced two new models on June 30 — Nano Banana 2 Lite, an image generation model, and Gemini Omni Flash, a video generation and editing model — per the company's announcement, with coverage the same day from Ars Technica, TechCrunch and independent AI researcher Simon Willison.

Nano Banana 2 Lite, carried in the API as gemini-3.1-flash-lite-image, is described by Google as its fastest and cheapest Gemini image model, engineered for velocity and scale: text-to-image generation in four seconds, priced at $0.034 per 1,000 images at 1K resolution, per the announcement. Google says the model targets rapid ideation and high-velocity workflows while retaining prompt adherence, character consistency and legible text rendering.

Gemini Omni Flash, available as gemini-omni-flash-preview, handles video generation and conversational editing via natural language, accepts text, image and video references, and produces 10-second generations, with longer durations to come, per Google. It is priced at $0.10 per second of video output. Both models are available in Google AI Studio, the Gemini API and the Gemini Enterprise Agent Platform, per the announcement, with Nano Banana 2 Lite also rolling out across Search's AI Mode, the Gemini app, Photos, NotebookLM and Google Ads.

Willison, testing the image model against a where's-Waldo-style prompt, wrote that he preferred its output to results from earlier Nano Banana models in April — while noting it spelled "Forest Festival" wrong in two different ways, a reminder that text rendering remains imperfect. Google's own notes flag limits on Omni Flash, including video references that are accepted up to 3 seconds but not properly processed and character consistency issues across scene changes, per the announcement.

Background

Nano Banana began as the internal codename for a Gemini image editing model that went viral in 2025 and became one of Google's most effective consumer AI hooks; the company has since kept the name as consumer-facing branding for its image model line. The Lite release follows a standard playbook across frontier labs: after a flagship model establishes quality, a smaller, faster variant follows to capture high-volume, price-sensitive usage — the tier where image generation is embedded into products at scale rather than invoked one image at a time.

What comes next

Omni Flash carries a preview designation, and Google says longer video durations are coming; watch for the model's move to general availability and whether the flagged reference-handling and consistency limits are resolved. The pricing — fractions of a cent per image, ten cents per second of video — sets a marker that competing labs' next releases will be measured against.