Nano Banana Pro is Google’s Gemini 3 Pro Image model. For the native Gemini API, its model ID is gemini-3-pro-image. Google’s current model reference lists it as Stable and Paid. Those labels confirm the model’s status in the reference, but do not establish your account’s eligibility, regional access, or project quotas. [1]
Make a first image-generation request in Python
You’ll need a Gemini API key, the Google Gen AI Python SDK, and billing configured for paid use. Google’s setup guide says paid-tier activation requires Cloud Billing; check the current billing flow for the amount and payment steps that apply to your account. Store the key in an environment variable, not in your source code. The SDK automatically reads GEMINI_API_KEY when creating a client without an explicit key. [2][3]
Install the SDK and set the key in your shell:
python -m pip install --upgrade google-genai pillow
export GEMINI_API_KEY="YOUR_API_KEY"
Save the following as generate.py, then run python generate.py:
from google import genai
client = genai.Client()
response = client.models.generate_content(
model="gemini-3-pro-image",
contents=(
"Create a polished product photo of a ceramic coffee mug on a desk, "
"soft morning light, clean neutral background."
),
)
parts = response.parts or []
saved = False
for part in parts:
if part.inline_data is not None:
part.as_image().save("nano-banana-pro.png")
saved = True
if not saved:
raise RuntimeError(
"The response contained no image part; inspect the API response "
"and any safety or generation feedback."
)
print("Saved nano-banana-pro.png")
This follows Google’s documented generate_content image-generation pattern, uses the model ID in its current reference, and saves returned image data using the SDK’s as_image() helper. The example has not been independently executed for this article. If it returns no image, inspect the response and its feedback rather than assuming generation succeeded. [1][4]
This example is for the Gemini Developer API. Google also lists Gemini 3 Pro Image for its enterprise platform; that route has separate project and authentication setup. Do not assume the API-key configuration above applies there. [3][5]
Production checks
Before integrating the call into an application, verify the requirements that affect your workload:
- Model limits: Google’s Gemini API model reference lists a 65,536-token input limit and a 32,768-token output limit. [1]
- Image options: Google’s AI Studio model comparison lists Nano Banana Pro output at 1K, 2K, and 4K. Test the resolution and aspect ratios your application needs. [6]
- Image inputs and grounding: The model reference lists image input and Google Search grounding as supported. Confirm the current image-input method and grounding configuration in Google’s image-generation guide before implementing either. [1][4]
- Quotas and access: Stable and Paid do not tell you your project’s request quotas, regional availability, or eligibility. Check these for the account and route you plan to deploy; don’t infer them from the model-status labels. [1]
- Output review: Google describes high-fidelity generation, text rendering, and image editing as model capabilities. Evaluate representative outputs for accuracy, legibility, and brand requirements before serving them to users. Google also says generated images include SynthID watermarks. [7]
Pricing caveats and model choice
Google lists Gemini 3 Pro Image as a Paid model. Check the official Gemini API pricing page for the current rate and billing unit before estimating spend; this article does not quote a numeric price. Do not treat a third-party wrapper’s “free” offer as free use of Google’s native API. [1][8]
Google’s launch article gave a qualitative comparison: Gemini 2.5 Flash Image for faster, lower-cost image generation, versus Gemini 3 Pro Image for higher quality at higher cost and latency. It did not provide measured latency or a numeric price comparison, so benchmark your own workload and consult current pricing before choosing. [7]
Sources
[1] Google, Gemini 3 Pro Image model reference. [2] Google AI Studio, Gemini API getting started. [3] Google Gen AI Python SDK documentation. [4] Google, Gemini image generation guide. [5] Google Cloud, Gemini 3 Pro Image model reference. [6] Google AI Studio, Nano Banana model comparison. [7] Google, Nano Banana Pro developer announcement. [8] Google AI Studio, Gemini Developer API pricing.
Sources
Gemini 3 Pro image | Gemini API | Google AI for Developers
Getting started | Gemini API | Google AI Studio
Google Gen AI SDK documentation
Nano Banana image generation | Gemini API
Gemini 3 Pro Image (Nano Banana Pro) | Gemini Enterprise Agent Platform
Nano Banana | Google AI Studio
Developers can build with Nano Banana Pro (Gemini 3 Pro Image)