Nano Banana Pro
google/nano-banana-proPosters and pro visuals with legible text inside the image
Nano Banana Pro is Google's top image model, launched on 20 November 2025 on top of Gemini 3 Pro. Google presents it as a design engine for complex visuals, precise layouts and text inside the image.
- Modalities
- Price / image
- 284cr
- Max resolution
- 2048x2048
- Released
- Nov 20, 2025
It is the pick for posters, packaging and campaign visuals where every word has to stay legible, in several languages according to Google. It accepts reference images so a product or a face stays consistent from one image to the next.
Through Tchavi you generate with the OpenAI Images API format, one key for 80+ text, image and video models, and you pay with mobile money, no card. Your credits never expire.
What this model does well
Text inside the image
Google highlights crisp, multilingual text built into the visual
Gemini 3 Pro core
Built on Gemini 3 Pro to follow complex instructions
Reference images
Keep a product or a face consistent from image to image
1K and 2K resolutions
Outputs at 1024×1024 or 2048×2048
Varied styles
Photo, illustration, art, 3D, logo, infographic
Example prompts
Professional flyer for a grill restaurant in Cotonou, warm colours. Large title "BRAISE & CO", subtitle "Open every evening from 6 pm", a plate of charcoal-grilled chicken in the foreground.
Minimalist logo for a tech startup called "Kora", simple geometric shape, two colours, light background, the name legible underneath.
Cinematic portrait of a woman, golden late-afternoon light, shallow depth of field, natural skin texture, eyes to the camera.
Good to know
- Every image carries Google's invisible SynthID watermark.
- Limited to 20 requests per minute: plan a queue for bulk generation.
- For everyday visuals at volume, Nano Banana is lighter; Pro is made for the final visual.
Frequently asked questions
Nano Banana Pro or Nano Banana 2: which one?
Nano Banana 2 is the balanced one, for a visual that will be published. Nano Banana Pro is the top tier: text inside the image, complex layouts, detailed instructions. If the text in the image is the critical part, pick Pro.
Nano Banana Pro or Imagen 4 Ultra: which one?
Imagen 4 Ultra aims at maximum photoreal fidelity, without reference images. Nano Banana Pro fits better when the image contains text or a layout, or when you need reference images.
Can it write French text inside the image?
Yes: Google highlights legible text in several languages. Put the exact text in quotes in the prompt, accents included, and proofread before printing.
Can it keep the same product or face?
Yes: send reference images with reference_images (URL or base64). According to Google, the model can blend up to 14 images; test the limit on your own cases.
How do I call Nano Banana Pro through Tchavi?
POST /v1/images/generations with model: "nano-banana-pro", your prompt and size (1024x1024 or 2048x2048). The OpenAI Images SDK works by changing the base URL; /v1/images/edits retouches.
Can I pay with mobile money, without a bank card?
Yes. Top up your Tchavi account with Orange Money, MTN MoMo, M-Pesa or another mobile money operator. No bank card is needed. Credits land as soon as the payment is confirmed and are deducted as you use them.
Is the API compatible with OpenAI's?
Yes. Keep your current OpenAI SDK, just change the base URL to Tchavi's (shown in the API tab) and use your Tchavi key. The rest of your code stays the same.
API
import Tchavi from '@tchavi/sdk';
const client = new Tchavi({ apiKey: 'API_KEY' });
const result = await client.images.generations.create({
"model": "nano-banana-pro",
"prompt": "A cute cat wearing a space suit",
"n": 1,
"quality": "standard",
"size": "1024x1024"
});
console.log(result.data[0].b64_json);
console.log('Credits used:', result.tchavi.credits_used);API parameters
Every parameter this model accepts on the API. Check the SDK for full typings.
modelRequiredID of the image model (e.g. `nano-banana-pro`).
promptRequiredText description of the image to generate.
Show all parameters (3)
n1Number of images to generate (1–4).