Skip to content
In Active Development

AI Image Generation
Inside Your Workflow

Generate images via API, SDK, CLI, Lab, or live URLs. Production-ready CDN delivery in seconds.

Free early access. No credit card required.

API-FirstBuilt-in CDNWeb LabStyle ReferencesPrompt URLs

One request: a production-ready URL

Plain REST API, nothing to install

# Generate an image
curl -X POST https://api.banatie.app/v1/generate \
-H "Authorization: Bearer $API_KEY" \
-d '{"prompt": "modern office interior, natural light"}'
# Response
{
"url": "https://cdn.banatie.app/img/a7x2k9.png",
"enhanced_prompt": "A photorealistic modern office...",
"generation_time": 12.4
}

CDN-cached and ready to use. No download step, no re-upload.

Why developers choose Banatie

The image pipeline is not your product: Banatie takes that part.

Placeholder hell

"I'll add images later" never happens

Generate real images as you build

Context switching

Leave IDE, generate elsewhere, come back

Stay in your workflow. API, SDK, MCP

Asset management

Download, optimize, upload, get URL

Production CDN URLs instantly

Style drift

Every image looks different

Reference images keep style consistent
Unique

Prompt URLs: images via HTML

Put a prompt in your img src and get a real image, no API calls and no JavaScript involved.

<!-- Write this -->
<img src="https://cdn.banatie.app/gen?p=modern office interior" />

<!-- Get this: production-ready image, cached, CDN-delivered -->

Perfect for placeholder images, prototypes, static sites, and AI coding agents.

Your prompt, your control, production-ready output

Prompt enhancement and CDN delivery are part of the same flow: one request in, one URL out.

1

Your Prompt

"a cat on windowsill"

2

Smart Enhancement

Style + details added

3

AI Generation

Gemini creates image

4

CDN Delivery

Instant global URL

What you control

Style templates : photorealistic, illustration, minimalist, etc
Reference images : @aliases keep visual consistency
Output specs : aspect ratio, dimensions, format

Enhanced prompts are visible in API response. You always see what was generated.

Built for real development workflows

Generation, references, flows, and delivery: the parts you would otherwise build yourself.

Reference Images

Use @aliases to maintain style consistency across your project. Reference up to 3 images per generation.

Flows

Chain generations, iterate on results, build image sequences with @last and @first references.

7 Style Templates

Same prompt, different styles: photorealistic, illustration, minimalist, product, comic, sticker.

Instant CDN Delivery

Every image gets production-ready URL. No upload step, no hosting setup.

Output Control

Aspect ratio, dimensions, format: set them per request. Square thumbnails or ultra-wide banners, same API.

Prompt URLs

Unique

Generate images via URL parameters. Put prompt in img src, get real image. Built-in caching.

Works with your tools

Use what fits your workflow. All methods, same capabilities.

REST API
TypeScript SDK
MCP Server
CLI
Banatie Lab
Prompt URLs

Banatie Lab: the official web interface for Banatie API. Generate images, build flows, browse your gallery, grab ready-to-use code snippets.

Perfect for Claude Code, Cursor, and any AI-powered workflow.

Shape the future of Banatie

We're building this for developers like you. Early adopters get direct influence on the roadmap: suggest features, vote on priorities.

Direct feedback channelFeature votingEarly adopter community

Powered by Google Gemini

We chose Gemini because it's the only model family that combines native multimodal understanding with production-grade image generation. Two models, optimized for different needs.

Gemini 2.5 Flash Image

Nano Banana

Optimized for speed and iteration. Perfect for rapid prototyping and high-volume generation.

  • Sub-3 second generation time
  • Multi-turn editing : refine through conversation
  • Up to 3 reference images for consistency
  • 1024px resolution output

Gemini 3 Pro Image

Nano Banana Pro

Maximum quality and creative control. For production assets and professional workflows.

  • Up to 4K resolution output
  • 14 reference images for brand consistency
  • Studio controls : lighting, focus, color grading
  • Thinking mode : advanced reasoning for complex prompts

Why Gemini outperforms competitors

Perfect Text Rendering

Legible text in images: logos, diagrams, posters. Other models still struggle with that.

Native Multimodal

Understands text AND images in one model. Not a text model + image model bolted together.

Precise Prompt Following

What you ask is what you get. No artistic "interpretation" that ignores your instructions.

Professional Realism

Photorealistic output that replaces stock photos. Real, usable images.

Gemini 2.5 Flash Image ranked #1 on LMArena for both text-to-image and image editing (August 2025)

Ready to build?

Enter your email, run one command from your inbox, and start generating images in minutes.

No credit card required • Free to start • Cancel anytime