What exactly is Nano Banana 2 and what does it do?

Nano Banana 2 is Google’s latest AI image generation and editing model, built on Gemini Flash image technology to deliver fast, high-quality visual generation and precise instruction following across text and image inputs.

How does Nano Banana 2 relate to Gemini 3.1 Flash Image?

Nano Banana 2 is essentially the consumer-facing branding for Google’s Gemini 3.1 Flash Image model, combining advanced capabilities from previous Nano Banana versions with the speed of Flash models.

What improvements does Nano Banana 2 add over earlier Nano Banana models?

Nano Banana 2 brings faster generation speed, sharper detail, better instruction fidelity, enhanced text rendering/localized translation, and broader creative control while making many Pro-grade features available at base tier.

What kinds of images and resolutions can Nano Banana 2 generate?

The model supports flexible output with various aspect ratios and resolutions up to 4K, suitable for social media, ads, displays, and professional content.

Can Nano Banana 2 maintain consistency in complex compositions?

Yes — it preserves consistency across multiple subjects and objects (e.g., up to five characters and 14 objects in a single prompt workflow), helping with narrative scenes and storyboard-style tasks.

What image generation use cases is Gemini 3.1 Flash Image best suited for?

It’s well-suited for professional-grade image creation and editing, infographics, multi-image consistency, text rendering, and localized multilingual outputs, especially when workflows need precise control and repeated iterations.

Does Nano Banana 2 use real-time information or world knowledge?

Nano Banana 2 incorporates real-world knowledge and image search integration to help generate more accurate subjects, infographics, and location-aware visuals.

Can Gemini 3.1 Flash Image generate detailed text within images or diagrams?

Yes — it can generate and render clear text within images, but extremely small or dense multi-paragraph text sometimes remains challenging.

Доступный API Nano Banana 2 | text-to-image

Playground для Nano Banana 2

Изучите Playground Nano Banana 2 — интерактивную среду для тестирования моделей и выполнения запросов в реальном времени. Попробуйте промпты, настройте параметры и итерируйте мгновенно, чтобы ускорить разработку и проверить варианты использования.

Technical Specifications of Gemini 3.1 Flash Image Preview

Item	Gemini 3.1 Flash Image Preview
Provider	Google
Model family	Gemini 3.1 (Flash tier)
Primary focus	Fast multimodal generation with image preview
Input types	Text, Image
Output types	Text, Image (preview generation)
Context window	Up to 1M tokens (Gemini 3.x Flash tier standard)
Latency tier	Low-latency, high-throughput
Streaming support	Yes
Tool calling	Yes (Gemini API tools framework)
Version	3.1

What is Nano Banana 2

Nano Banana 2 is the popular nickname used by the press and developer community for the newly released Gemini-3.1-Flash-Image model. Google positions it as the “Flash”-tier image engine that brings near-Pro visual fidelity to a much lower latency and cost tier — suitable for high-volume generation, rapid iterative editing, and integrated product workflows across Google services. It inherits Gemini 3.1’s multimodal reasoning and adds image-centric capabilities (legible text in images, multi-image composition, wide aspect ratio support, native 4K).

Main features

High-speed, multi-resolution generation: Flash-tier speed with options for 0.5K / 1K / 2K / 4K outputs and new extreme aspect ratios (1:4, 4:1, 1:8, 8:1).
Real-time web grounding: Integrates both text and image search results to ground generated content in current web information when “Thinking” or search grounding is enabled. Useful for up-to-date references and factual infographics.
Improved text rendering: Better short-text and graphic text rendering (fonts, sizes) than earlier Flash models; still imperfect on long paragraphs/small text.
Multi-input editing and multi-turn workflows: Strong support for combining several images as inputs and for iterative edits across turns.

📊 Benchmark Performance — Image Generation & Editing (Elo scores)

Capability	Gemini 3.1 Flash Image (Nano Banana 2)	Gemini 2.5 Flash Image (Nano Banana)	Gemini 3 Pro Image (Nano Banana Pro)	GPT-Image 1.5	Seedream 5.0 Lite	Grok Imagine Image Pro
Text-to-Image — Overall Preference	1079.0 ± 7.0	1073.0 ± 5.0	942.0 ± 6.0	1021.0 ± 5.0	1047.0 ± 5.0	928.0 ± 8.0
Text-to-Image — Visual Quality	1140.0 ± 6.0	1129.0 ± 6.0	929.0 ± 6.0	1043.0 ± 5.0	975.0 ± 5.0	759.0 ± 10.0
Text-to-Image — Infographics (Factuality)	1114.0 ± 14.0	1074.0 ± 12.0	881.0 ± 13.0	1102.0 ± 13.0	985.0 ± 12.0	890.0 ± 22.0
Editing — General	1065.0 ± 9.0	1047.0 ± 9.0	913.0 ± 9.0	1051.0 ± 10.0	995.0 ± 8.0	937.0 ± 9.0
Editing — Character	1056.0 ± 7.0	1049.0 ± 7.0	952.0 ± 7.0	1050.0 ± 8.0	1025.0 ± 7.0	894.0 ± 8.0
Editing — Creative	1023.0 ± 7.0	1031.0 ± 7.0	976.0 ± 7.0	1004.0 ± 7.0	1017.0 ± 7.0	938.0 ± 7.0
Editing — Object/Environment	1029.0 ± 8.0	1018.0 ± 8.0	945.0 ± 8.0	1042.0 ± 10.0	976.0 ± 8.0	946.0 ± 9.0
Editing — Multi-Input	1037.0 ± 8.0	1016.0 ± 8.0	919.0 ± 9.0	1056.0 ± 12.0	1014.0 ± 9.0	N/A
Editing — Stylization	1045.0 ± 7.0	1031.0 ± 7.0	862.0 ± 8.0	1045.0 ± 9.0	996.0 ± 7.0	984.0 ± 7.0

Key takeaways from this benchmark table:

Across text-to-image generation and image editing categories, Gemini 3.1 Flash Image consistently leads or matches the highest scores among Flash-tier and many competitive image models.
The model shows especially strong results in Visual Quality and Infographic (Factuality) benchmarks—signaling that it excels not only in aesthetic quality but also in rendering structurally accurate content.
On Multi-Input editing, Nano Banana 2 also shows robust generalization, with higher scores than its previous Flash generation.

These evaluations are conducted via human side-by-side Elo comparisons on a diverse benchmark suite, reflecting both preference and fidelity across commonly used image generation/editing tasks.

Nano Banana 2 vs Nano Banana vs Nano Banana Pro

Model	Positioning	Representative benchmark/notes
Gemini 3.1 Flash Image (Nano Banana 2)	Flash tier: speed + high visual quality (2K–4K)	Overall preference 1079.0 ± 7.0; visual quality 1140 ± 6.0 (internal GenAI-Bench).
Gemini 2.5 Flash Image (Nano Banana)	Earlier Flash release (lower fidelity)	Slightly lower preference/visual scores vs 3.1.
Gemini 3 Pro Image (Nano Banana Pro)	Pro tier: higher perceived fidelity for complex tasks, higher cost/latency	Different tradeoffs; some metrics show different relative rankings in specialty tasks.
GPT-Image 1.5 / other commercial models	Competitors (open/closed)	In Google’s internal benchmarks GPT-Image and others scored below Gemini 3.1 on visual quality and overall preference in the reported eval. Independent third-party comparisons vary.

When to choose Flash Image Preview:

Real-time image preview in apps
Cost-sensitive large-scale image generation
Interactive design assistants

How to access and integrate Nano Banana 2

Log in to cometapi.com. If you are not our user yet, please register first. Sign into your CometAPI console. Get the access credential API key of the interface. Click “Add Token” at the API token in the personal center, get the token key: sk-xxxxx and submit.

Step 2: Send Requests to `Nano Banana 2` API

Select the “gemini-3.1-flash-image-preview8” endpoint to send the API request and set the request body. The request method and request body are obtained from our website API doc. Our website also provides Apifox test for your convenience. Replace <YOUR_API_KEY> with your actual CometAPI key from your account. Where to call it:Gemini generates image

Nano Banana 2 supports image editing, image generation, and multi-image workflows. For image editing, you need to upload the image URL. For more parameters, please refer to the documentation.

Step 3: Retrieve and Verify Results

Process the API response to get the generated answer. After processing, the API responds with the task status and output data. You can directly download the image to your local machine in the playground (usually in PNG format). An image URL is generated in the API process; please download it promptly.

Цены для Nano Banana 2

Изучите конкурентоспособные цены на Nano Banana 2, разработанные для различных бюджетов и потребностей использования. Наши гибкие планы гарантируют, что вы платите только за то, что используете, что упрощает масштабирование по мере роста ваших требований. Узнайте, как Nano Banana 2 может улучшить ваши проекты, сохраняя при этом управляемые расходы.

nano-banana-2（image）

variant / alias	Price
gemini-3.1-flash-image (0.5K)	≈ $0.03600
gemini-3.1-flash-image (1K)	≈ $0.05360
gemini-3.1-flash-image (2K)	≈ $0.08080
gemini-3.1-flash-image (4K)	≈ $0.12080
gemini-3.1-flash-image-preview (0.5K)	≈ $0.03600
gemini-3.1-flash-image-preview (1K)	≈ $0.05360
gemini-3.1-flash-image-preview (2K)	≈ $0.08080
gemini-3.1-flash-image-preview (4K)	≈ $0.12080

Пример кода и API для Nano Banana 2

Получите доступ к исчерпывающим примерам кода и ресурсам API для Nano Banana 2, чтобы упростить процесс интеграции. Наша подробная документация предоставляет пошаговые инструкции, помогая вам использовать весь потенциал Nano Banana 2 в ваших проектах.

# Get your CometAPI key from https://api.cometapi.com/console/token
# Export it as: export COMETAPI_KEY="your-key-here"

mkdir -p ./output

curl -s "https://api.cometapi.com/v1beta/models/gemini-3.1-flash-image-preview:generateContent" \
  -H "Authorization: $COMETAPI_KEY" \
  -H 'Content-Type: application/json' \
  -X POST \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "A woman leaning on a wooden railing of a traditional Chinese building. She is wearing a blue cheongsam with pink and red floral motifs and a headdress made of colorful flowers, including roses and lilacs. Realistic painting style, focusing on the textural details of the clothing patterns and wooden buildings."
          }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": {
        "aspectRatio": "9:16"
      }
    }
  }' | python3 -c "
import sys, json, base64
data = json.load(sys.stdin)
parts = data['candidates'][0]['content']['parts']
for part in parts:
    if 'text' in part:
        print(part['text'])
    elif 'inlineData' in part:
        img = base64.b64decode(part['inlineData']['data'])
        with open('./output/gemini-3.1-flash-image-preview.png', 'wb') as f:
            f.write(img)
        print('Image saved to ./output/gemini-3.1-flash-image-preview.png')
"

cURL Code Example

# Get your CometAPI key from https://api.cometapi.com/console/token
# Export it as: export COMETAPI_KEY="your-key-here"

mkdir -p ./output

curl -s "https://api.cometapi.com/v1beta/models/gemini-3.1-flash-image-preview:generateContent" \
  -H "Authorization: $COMETAPI_KEY" \
  -H 'Content-Type: application/json' \
  -X POST \
  -d '{
    "contents": [
      {
        "role": "user",
        "parts": [
          {
            "text": "A woman leaning on a wooden railing of a traditional Chinese building. She is wearing a blue cheongsam with pink and red floral motifs and a headdress made of colorful flowers, including roses and lilacs. Realistic painting style, focusing on the textural details of the clothing patterns and wooden buildings."
          }
        ]
      }
    ],
    "generationConfig": {
      "responseModalities": ["IMAGE"],
      "imageConfig": {
        "aspectRatio": "9:16"
      }
    }
  }' | python3 -c "
import sys, json, base64
data = json.load(sys.stdin)
parts = data['candidates'][0]['content']['parts']
for part in parts:
    if 'text' in part:
        print(part['text'])
    elif 'inlineData' in part:
        img = base64.b64decode(part['inlineData']['data'])
        with open('./output/gemini-3.1-flash-image-preview.png', 'wb') as f:
            f.write(img)
        print('Image saved to ./output/gemini-3.1-flash-image-preview.png')
"

Python Code Example

from google import genai
from google.genai import types
from PIL import Image
import os

# Get your CometAPI key from https://api.cometapi.com/console/token, and paste it here
COMETAPI_KEY = os.environ.get("COMETAPI_KEY") or "<YOUR_COMETAPI_KEY>"
BASE_URL = "https://api.cometapi.com"

client = genai.Client(
    http_options={"api_version": "v1beta", "base_url": BASE_URL},
    api_key=COMETAPI_KEY,
)

prompt = (
    "A woman leaning on a wooden railing of a traditional Chinese building. "
    "She is wearing a blue cheongsam with pink and red floral motifs and a headdress "
    "made of colorful flowers, including roses and lilacs. Realistic painting style, "
    "focusing on the textural details of the clothing patterns and wooden buildings."
)
aspect_ratio = "9:16"  # "1:1","2:3","3:2","3:4","4:3","4:5","5:4","9:16","16:9","21:9"

response = client.models.generate_content(
    model="gemini-3.1-flash-image-preview",
    contents=[prompt],
    config=types.GenerateContentConfig(
        response_modalities=["IMAGE"],
        image_config=types.ImageConfig(aspect_ratio=aspect_ratio),
    ),
)

os.makedirs("./output", exist_ok=True)

for part in response.parts:
    if part.text is not None:
        print(part.text)
    elif part.inline_data is not None:
        image = part.as_image()
        output_path = "./output/gemini-3.1-flash-image-preview.png"
        image.save(output_path)
        print(f"Image saved to {output_path}")

JavaScript Code Example

import fs from "fs";
import path from "path";

// Get your CometAPI key from https://api.cometapi.com/console/token, and paste it here
const api_key = process.env.COMETAPI_KEY || "<YOUR_COMETAPI_KEY>";
const base_url = "https://api.cometapi.com/v1beta";
const model = "gemini-3.1-flash-image-preview";

const prompt =
  "A woman leaning on a wooden railing of a traditional Chinese building. " +
  "She is wearing a blue cheongsam with pink and red floral motifs and a headdress " +
  "made of colorful flowers, including roses and lilacs. Realistic painting style, " +
  "focusing on the textural details of the clothing patterns and wooden buildings.";

const response = await fetch(`${base_url}/models/${model}:generateContent`, {
  method: "POST",
  headers: {
    "Content-Type": "application/json",
    Authorization: api_key,
  },
  body: JSON.stringify({
    contents: [
      {
        role: "user",
        parts: [{ text: prompt }],
      },
    ],
    generationConfig: {
      responseModalities: ["IMAGE"],
      imageConfig: {
        aspectRatio: "9:16",
      },
    },
  }),
});

const data = await response.json();

const outputDir = "./output";
if (!fs.existsSync(outputDir)) {
  fs.mkdirSync(outputDir, { recursive: true });
}

for (const candidate of data.candidates) {
  for (const part of candidate.content.parts) {
    if (part.text) {
      console.log(part.text);
    } else if (part.inlineData) {
      const imageBuffer = Buffer.from(part.inlineData.data, "base64");
      const outputPath = path.join(outputDir, "gemini-3.1-flash-image-preview.png");
      fs.writeFileSync(outputPath, imageBuffer);
      console.log(`Image saved to ${outputPath}`);
    }
  }
}

Версии Nano Banana 2

Причина наличия нескольких снимков Nano Banana 2 может включать такие потенциальные факторы, как: изменения в выходных данных после обновлений, требующие сохранения старых снимков для обеспечения согласованности; предоставление разработчикам переходного периода для адаптации и миграции; а также наличие разных снимков, соответствующих глобальным или региональным конечным точкам для оптимизации пользовательского опыта. Для получения подробной информации о различиях между версиями обратитесь к официальной документации.

Model id	description	Availability	Request
gemini-3.1-flash-image	Recommend, Pointing to the latest model	✅	Gemini generates image
gemini-3.1-flash-image-preview	Official Preview	✅	Gemini generates image

Nano Banana 2