nano banana 2 lite · nano-banana Reference

Nano Banana 2 Lite (`gemini-3.1-flash-lite-image`) Model Card #

Overview #

Nano Banana 2 Lite (gemini-3.1-flash-lite-image) is Google’s ultra-high-velocity, cost-efficient image generation model designed for real-time consumer applications, conversational image editing, and high-volume generation pipelines.

Model Code: gemini-3.1-flash-lite-image
Release Date: June 2026
Status: General Availability (GA)
DeepMind Model Card: gemini-3-1-flash-lite-image ↗

Technical Specifications #

Property	Value
Input Modalities	Text, Image / PDF
Output Modalities	Image and Text (Interleaved)
Input Token Limit	65,536 tokens
Output Token Limit	4,096 tokens
Max Resolution	1K only (1024px bounding box; 2K/4K unsupported)
Generation Latency	Sub-2 seconds (< 2s) end-to-end
Function Calling	Supported
Thinking Mode	Supported (`minimal` [default], `high`)
Search Grounding	❌ Not Supported
Watermarking	SynthID (Always On) + C2PA Metadata

Supported Aspect Ratios & Resolutions (1K Fixed) #

Nano Banana 2 Lite generates images within a 1K bounding box across 14 discrete aspect ratios:

Aspect Ratio	Output Resolution	Description / Best For
`1:1`	1024 × 1024	Standard square, avatars, profile icons
`2:3`	848 × 1264	Classic portrait photography
`3:2`	1264 × 848	Classic landscape photography
`3:4`	896 × 1200	Vertical mobile screens, social stories
`4:3`	1200 × 896	Standard landscape monitor display
`4:5`	928 × 1152	Social media feed portrait
`5:4`	1152 × 928	Medium format display
`9:16`	768 × 1376	Vertical video stills, mobile wallpaper
`16:9`	1376 × 768	Widescreen presentations, YouTube thumbnails
`21:9`	1584 × 672	Ultrawide cinematic banners
`1:4`	512 × 2048	Tall vertical bookmarks and sidebars
`4:1`	2048 × 512	Wide horizontal headers and website banners
`1:8`	384 × 3072	Extreme tall vertical infographics
`8:1`	3072 × 384	Extreme wide horizontal panorama headers

Capabilities & Limitations #

Strengths #

Ultra-Low Latency: Optimized for real-time interaction, UI mockups, stickers, and fast conversational iteration.
Cost Efficiency: Lowest compute and token consumption in the Nano Banana family.
Conversational Multi-Turn Edits: Fast local modifications (e.g. “change hair color to blue”, “add sunglasses”).
Function Calling: Fully integrated with tool-calling workflows.

Limitations #

Resolution Capped at 1K: Cannot generate 2K (2048px) or 4K (4096px) images. Use Nano Banana 2 or Nano Banana Pro for 2K/4K.
No Search Grounding: Does not support real-time web or image search grounding tools.
Single/Local Reference Focus: Best for single-image edits; not optimized for complex multi-character cross-scene consistency.

API Code Examples #

1. Python SDK (`google-genai`) - Interactions API #

python

import base64
from google import genai

client = genai.Client()

# Ultra-fast text-to-image generation
interaction = client.interactions.create(
    model="gemini-3.1-flash-lite-image",
    input="A vibrant sticker of a cute robot drinking bubble tea, white background",
    response_format={
        "type": "image",
        "aspect_ratio": "1:1",
        "image_size": "1K",
    },
)

if interaction.output_image:
    with open("robot_sticker.png", "wb") as f:
        f.write(base64.b64decode(interaction.output_image.data))

2. Python SDK (`google-genai`) - Models API (`generate_content`) #

python

from google import genai
from PIL import Image

client = genai.Client()

response = client.models.generate_content(
    model="gemini-3.1-flash-lite-image",
    contents="A kawaii watercolor illustration of a sleeping kitten in a teacup",
)

for part in response.candidates[0].content.parts:
    if part.inline_data:
        with open("kitten.png", "wb") as f:
            f.write(part.inline_data.data)
        break

3. JavaScript / TypeScript SDK (`@google/genai`) #

typescript

import { GoogleGenAI } from "@google/genai";
import * as fs from "node:fs";

const ai = new GoogleGenAI({});

async function generate() {
  const interaction = await ai.interactions.create({
    model: "gemini-3.1-flash-lite-image",
    input: "Minimalist geometric icon of a rocket launching",
    response_format: {
      type: "image",
      aspect_ratio: "1:1",
      image_size: "1K",
    },
  });

  if (interaction.output_image) {
    fs.writeFileSync("rocket.png", Buffer.from(interaction.output_image.data, "base64"));
  }
}

generate();

4. REST / cURL #

bash

curl -s -X POST \
  "https://generativelanguage.googleapis.com/v1beta/interactions" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gemini-3.1-flash-lite-image",
    "input": "A clean flat design icon of a coffee cup with steam",
    "response_format": {
      "type": "image",
      "aspect_ratio": "1:1",
      "image_size": "1K"
    }
  }'

Nano Banana 2 Lite (gemini-3.1-flash-lite-image) Model Card #