Try Seedance 2.5 Uncensored Video from ByteDance →
Sign InStart Building

Task History

Click to see output list

No tasks yet

Go to Models
Image GenerationActive

google / nano-banana-2-lite

nano-banana-2-lite

bygoogle

Google’s Nano Banana 2 Lite generates and edits 1K images from text, with optional multi-image references. It’s built for rapid iteration and draft assets.

Text to ImageImage to ImageImage EditingFast Inference
Model ID
nano-banana-2-lite
Provider
google
Updated
1783092421
wiro playground—google/nano-banana-2-lite
Reset to defaults
Delete All
0 / 14
Maximum 14 image allowed
Drop image to upload

OR

Click to browse your device

Supports: JPG, JPEG, PNG, GIF, WEBP, HEIC

Gemini 3.1 Flash Lite Image lets you mix up to 14 reference images of objects with high fidelity to include in the final image. Not optimized for character consistency or multi-turn sequential editing.

The prompt to generate the image.

Sample outputs
Sample 1
Sample 2
Sample 3
Sample 4
Click To See More Samples
Updated 1783092421

Overview

Nano Banana 2 Lite is Google’s Nano Banana image model tuned for efficient image generation and editing. It takes a text instruction, and it returns a single 1K image you can download. You can also provide reference images to guide composition, style, or targeted edits. It works well when you need many quick visual options without jumping to higher-resolution models.

What you can build

  • Rapid concept art and mood boards for games, film, or product direction
  • Marketing drafts like ad variants, banner concepts, and social thumbnails
  • Product mockups and packaging explorations with editable layouts
  • Style transfer experiments using multiple reference images
  • Background swaps, color changes, and object tweaks on an existing image
  • Multi-reference composition, like combining a product photo with a style reference

Inputs

  • A text instruction that describes the image you want, or the edit you want applied.
  • Optional reference images you upload, up to 14 images per request, to guide the output.
  • Reference images should be standard web image formats like JPEG, PNG, or WebP.
  • An output aspect ratio choice for the final image, such as square, portrait, landscape, or ultrawide.
  • A safety strictness setting that controls how aggressively the system blocks unsafe requests.

Outputs

The model returns one generated image.

  • File format: a raster image, typically JPEG or PNG.
  • Resolution class: fixed to 1K output.
  • Common 1K pixel sizes by aspect ratio: - 1:1 returns 1024 x 1024
    • 2:3 returns 848 x 1264
    • 3:2 returns 1264 x 848
    • 3:4 returns 896 x 1200
    • 4:3 returns 1200 x 896
    • 4:5 returns 928 x 1152
    • 5:4 returns 1152 x 928
    • 9:16 returns 768 x 1376
    • 16:9 returns 1376 x 768
    • 21:9 returns 1584 x 672

Limitations

  • Output resolution is limited to 1K. It can’t generate 2K or 4K images.
  • It’s designed for speed and throughput. It may lose fine detail compared with larger Nano Banana tiers.
  • It isn’t optimized for strong character consistency across many sequential edits.
  • It doesn’t support Google Search or Image Search grounding for real-time accuracy.
  • The model’s knowledge cutoff is January 2025, so it may miss newer facts.
  • Token limits apply: up to 65,536 input tokens and 4,096 output tokens.
  • Low-quality references can degrade results. Blur, compression artifacts, and bad lighting often carry through.
  • Mixed reference sets can confuse the model. Conflicting styles or perspectives may produce unexpected blends.

Safety & compliance

This model uses safety filters that may refuse prompts or edits that violate policy.

  • Follow Google’s Generative AI Prohibited Use Policy.
  • Don’t upload or request content that violates privacy or consent, including non-consensual intimate imagery.
  • Generated images include provenance and watermarking systems such as SynthID and C2PA content credentials.
  • If you see refusals, revise the prompt to remove unsafe or sensitive details.

Example prompts

Great starting points for nano-banana-2-lite.

An adorable fluffy baby owl wearing tiny steampunk goggles, perched on an antique brass telescope in a cozy dusty attic, warm golden hour sunlight streaming through a circular window, macro photography, hyper-detailed texture, depth of field.Image Generation
A wide-angle architectural photograph of a modern, minimalist living room at twilight. Floor-to-ceiling glass windows look out onto a dense, misty pine forest. Inside, the room features a low-profile charcoal grey linen sofa, a matte black steel coffee table, and a glowing warm-toned floor lamp casting a soft amber pool of light. The polished concrete floor subtlely reflects the ambient light from the lamp and the cool blue light from the twilight outside. Photorealistic, captured with a Canon EOS R5, 16mm lens, f/8, crisp edges, architectural digest style, perfectly balanced exposure.Image Generation
Extreme macro food photography of a freshly baked, steaming slice of cherry pie on a rustic, chipped ceramic plate. A scoop of melting vanilla bean ice cream sits on top, with tiny, distinct black vanilla specks visible in the cream. The pie features a flaky, golden-brown lattice crust with glistening, rich dark-red cherry filling slowly oozing onto the plate. Resting on a dark, heavily textured reclaimed oak table. Directional backlighting from a softbox highlights the wisps of steam and the glossy texture of the filling. Shot with a Sony A7R IV, 100mm macro lens, f/5.6, razor-sharp focus.Image Generation
A National Geographic style wildlife photograph of a red fox mid-stride, trotting through a dense, frost-covered meadow during the freezing early morning blue hour. Crisp, delicate white frost coats the individual blades of tall, dormant grass. The fox\'s thick winter coat is a vibrant, highly detailed orange-red, contrasting sharply against the muted, cold blue tones of the environment. A small puff of condensation is visible from the fox\'s breath in the freezing air. Captured with a telephoto 400mm lens, f/4, 1/1000s shutter speed, sharp focus directly on the fox\'s amber eyes, smooth bokeh background.Image Generation
A detailed, photorealistic photograph of the 1960s American four-door sedan from image, viewed from the exact same front three-quarter perspective on the gravel lot. The car is no longer derelict but has been completely and meticulously restored to factory-new condition. All surface rust is gone, replaced by a flawless, deep, lustrous coat of classic midnight-blue paint with a high-gloss finish. The chrome grille, front bumper, all trim, wheel covers, and window surrounds are polished to a mirror-like sheen, gleaming under the golden hour light (which remains soft and clean). The glass is crystal clear. The interior is pristine, featuring newly upholstered classic vinyl. The specific license plate is clear and legible. The overall composition and background elements, including the autumn-colored trees and power lines, remain identical, but from a pristine version of the scene.Image Generation

API quick start

Run nano-banana-2-lite with a single API call.

POST https://api.wiro.ai/v1/Run/google/nano-banana-2-lite
{
  "prompt": "A detailed, photorealistic photograph of …",
  "inputImage": "https://your-cdn.com/input.png",
  "aspectRatio": "auto",
  "safetySetting": "OFF"
}
View full API docs