Google Ships Nano Banana 2 Lite and Opens Gemini Omni Flash to Developers

Google has released Nano Banana 2 Lite, an image model that generates text-to-image output in 4 seconds at $0.034 per 1K-resolution image, and opened its video model Gemini Omni Flash to developers for the first time via Google AI Studio, the Gemini API and Gemini Enterprise Agent Platform.

The two releases land together by design. Google says developers can chain them: use Nano Banana 2 Lite for high-speed image generation, then pass the image as a reference to Gemini Omni Flash to animate it into video. The company frames the pairing as a way to build "comprehensive, end-to-end multimedia experiences that connect rapid image generation with video creation and editing," whether a workflow requires generating thousands of images or editing multi-turn video sequences.

The release matters for a generative media market where unit economics increasingly decide which models get embedded in production pipelines. At $0.034 per image, Nano Banana 2 Lite targets the high-volume tier that competitor image models have struggled to serve at low latency, and Omni Flash matches Veo 3.1 Fast's $0.10-per-second video pricing.

Nano Banana 2 Lite: speed and cost first

The model, identified in the API as gemini-3.1-flash-lite-image, is built for rapid ideation and high-velocity pipelines where speed and cost are the primary constraints. Google positions it as the recommended replacement for the original Nano Banana (gemini-2.5-flash-image), telling developers they can swap it out now for "immediate benefits across key performance dimensions." Google published benchmark comparisons of Nano Banana 2 and 2 Lite against competitor AI image models, evaluating Elo scores for generation and editing quality against latency and cost per image.

Despite the speed priority, Google says the model retains reliable prompt adherence, strong character consistency and legible in-image text rendering — a persistent weak point for image generators.

The company now organizes its image lineup in four tiers: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) for near-real-time, high-volume workflows; Nano Banana 2 (Gemini 3.1 Flash Image) as the "generalist workhorse" balancing quality and cost; Nano Banana Pro (Gemini 3 Pro Image) for professional use cases requiring the most robust control and advanced reasoning; and the legacy Nano Banana (Gemini 2.5 Flash Image), for which Google recommends upgrading.

Beyond developer platforms, Nano Banana 2 Lite is rolling out today across Google's consumer surfaces: AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow and Google Ads.

Gemini Omni Flash reaches developers

First shown at Google I/O, Gemini Omni Flash (gemini-omni-flash-preview) is now in public preview in Google AI Studio and the Gemini API. It natively supports video generation and conversational editing from combined text, image and video inputs. Google describes it as the model "where Gemini's multimodal reasoning meets video generation and editing."

Its capabilities include:

• Conversational video editing: refine and edit videos using natural language.
• Multimodal referencing: combine images, text and video inputs to maintain control and consistency over a scene.
• Real-world knowledge: the model draws on Gemini's knowledge of subjects like history, biology and narrative logic to construct videos.
• Text and action synchronization: connect text and graphics directly to video actions through simple prompting.

Google lists concrete limitations. Generations currently cap at 10 seconds, with longer durations promised. Uploading audio references and scene extension are not yet supported in the Gemini API. Video references up to 3 seconds are accepted by the API schema but are not correctly processed by the model. Character consistency during scene changes or panning movements has known issues that Google says it is working to improve. Full benchmarking data is available on Google DeepMind's Gemini Omni webpage.

Chaining the two models

Google is pushing developers toward multi-turn workflows. Using the Interactions API, developers can maintain session history and context so users can stack up to three sequential edits in a single session.

The company shipped three demo apps to illustrate the pattern. Anywhere takes a selfie or uploaded photo and uses Nano Banana 2 Lite to place the subject at dozens of iconic landmarks; clicking an image triggers Omni Flash to turn it into an animated clip of the location. Space Lift is an interior design app that generates fully realized room concepts across design aesthetics, then produces a cinematic video showcase of a chosen look. Omni product studio converts static images made by Nano Banana 2 Lite into cinematic e-commerce videos via Gemini Omni.

All three apps are available to remix.

Safety and provenance

Both models run on Google's infrastructure and apply SynthID watermarking to generated content. Users can verify AI content through the Gemini app, Gemini in Chrome or Search, part of Google's broader push to surface how content was created and edited across the web.

The pricing, latency and consumer-surface footprint of these releases signal Google's intent to make high-volume generative media a commodity layer inside its existing products — and to give developers an economic reason to build media pipelines entirely on Gemini infrastructure rather than stitching together models from multiple providers.