Nano Banana - Kyma API

Overview

This is Google’s Gemini-based image generation model. It accepts text and image inputs to produce images, with native support for edit-mode workflows where you pass an existing image and a prompt to generate a modified output. It offers three resolution tiers: 512, 1024, and 2048. On Kyma, it runs on the cheap tier with fast inference speeds and supports prompt caching for repeated prefixes. The endpoint is scheduled for retirement on 2026-10-20 as Google discontinues the underlying Gemini 2.5 endpoint; Kyma will automatically handle request failover if serving paths degrade, and responses will include exact costs in the usage.cost field. It does not support reasoning or structured outputs, and it is strictly an image-to-image or text-to-image generator.

Specs

Field	Value
Model ID	`nano-banana`
Best for	In-context image editing, style transfer, low-cost iteration
Context window	0
Input modalities	Text, Image
Output modalities	Image
Tool calling	Yes
Structured outputs	Yes
Prompt caching	Yes
Speed	Fast
Cost band	Cheap
Release stage	Stable

Pricing

	Per 1M tokens
Input	$0.20
Output	$0.60

Use this when

In-context image editing — Pass a source image with a text prompt to modify specific regions or apply new styles.
Low-cost asset iteration — Generate multiple visual drafts quickly while staying within a tight compute budget.
Style transfer workflows — Apply consistent visual themes across different base images using text-guided transformations.

Not ideal for

Do not use this model for tasks requiring structured JSON outputs, complex logical reasoning, or high-fidelity photorealism beyond its standard resolution tiers.

Pick something else when

You need a direct successor after the 2026-10-20 retirement: use nano-banana-3-flash.
You need higher resolution or premium photorealism: use imagen-4-ultra.
You need vector graphics or design-ready assets: use recraft-v4-vector-pro.

Example

from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="ky-...")

response = client.chat.completions.create(
    model="nano-banana",
    messages=[{"role": "user", "content": "..."}],
)

Agent query example

Ask the API which models fit, instead of hardcoding an id:

curl "https://kymaapi.com/v1/models?recommended_for=generate-images&tools=true&quality_tier=frontier-open"

FAQ

Will my existing API calls break when this model retires? The endpoint will remain active until October 20, 2026. After that date, route requests to nano-banana-3-flash, which is the designated replacement. Can I cache prompts to reduce costs? Yes, prompt caching is supported and bills repeated prompt prefixes at this model’s cached input rate. Does it support structured JSON or reasoning outputs? No, this model only outputs images and does not support reasoning or structured output formats.

​Overview

​Specs

​Pricing

​Use this when

​Not ideal for

​Pick something else when

​Example

​Agent query example

​FAQ