AI Image Generator CLI

Standalone Node.js CLI tool for generating and editing images using OpenAI (DALL-E 3 / gpt-image-1) and Google Gemini (Gemini Flash Image / Imagen 4). No npm install required — just Node.js and your API key.

---

Features

• Dual provider support — use OpenAI, Gemini, or both side by side
• Image generation — create images from text prompts
• Image editing — modify existing images with text descriptions
• Model selection — DALL-E 3 and gpt-image-1 (OpenAI), or auto-detected Gemini/Imagen models
• Multiple size options — DALL-E 3: 1024x1024, 1792x1024, 1024x1792 | gpt-image-1: 1024x1024, 512x512, 256x256 | Gemini: automatic
• Generation history — JSON log tracks every generation with prompt, timestamp, provider, model, and file path
• Smart model discovery — automatically finds image-capable models available on your Gemini API key
• Rate limit handling — detects 429 responses, parses retry delay, waits and retries automatically
• Zero dependencies — single .js file, uses only Node.js built-in modules
• Windows, macOS, Linux

---

Requirements

• Node.js 18+
• At least one API key:
  • OpenAI — platform.openai.com/api-keys
  • Gemini — ai.google.dev

Gemini note: Image generation requires a paid API key with billing enabled. Free tier keys have a quota of 0 for image generation. Enable billing at ai.google.dev — image models unlock automatically.

---

Quick Start

# OpenAI only
OPENAI_API_KEY=sk-your-key-here node image-gen.js

# Gemini only
GEMINI_API_KEY=AIza-your-key-here node image-gen.js

# Both providers
OPENAI_API_KEY=sk-abc123 GEMINI_API_KEY=AIza-xyz789 node image-gen.js

---

Setting API Keys

Linux / macOS:

export OPENAI_API_KEY=sk-abc123
export GEMINI_API_KEY=AIza-xyz789
node image-gen.js

Windows (Command Prompt):

set OPENAI_API_KEY=sk-abc123 && set GEMINI_API_KEY=AIza-xyz789 && node image-gen.js

Windows (PowerShell):

$env:OPENAI_API_KEY="sk-abc123"; $env:GEMINI_API_KEY="AIza-xyz789"; node image-gen.js

---

Menu

  ---- Menu ----
    1) Generate - Create an image from a text prompt
    2) Edit     - Edit an existing image with a text prompt
    3) History  - View past generations
    4) Quit

1) Generate

Choose provider, enter a prompt, select model and size (OpenAI only). Image is generated, saved, and logged.

  > A golden retriever wearing sunglasses on a beach at sunset

  Provider: OpenAI | Model: dall-e-3 | Size: 1024x1024
  Generating...

  Saved: image_output/generated_openai_2026-03-15T16-30-00.png (1247.3 KB)

2) Edit

Point to an existing image, describe your changes. Edited image saves as a new file.

  Image path: ./my-photo.png
  Edit prompt: Add a rainbow in the background

  Provider: OpenAI | Model: gpt-image-1 | Size: 1024x1024
  Editing...

  Saved: image_output/edited_openai_2026-03-15T16-35-00.png (982.1 KB)

Accepts PNG, JPG, JPEG, and WebP.

3) History

Lists all past generations and edits with prompt, provider, model, size, and file path.

---

Output

All images save to image_output/ in the current directory:

image_output/
  generated_openai_2026-03-15T16-30-00.png
  generated_gemini_2026-03-15T16-40-00.png
  edited_openai_2026-03-15T16-35-00.png
  history.json

Filenames follow the pattern: {type}_{provider}_{timestamp}.png

history.json stores a full log of every generation — prompt, provider, model, size, file path, and timestamp.

---

Supported Models

OpenAI

Model	Sizes	Notes
dall-e-3	1024x1024, 1792x1024, 1024x1792	Default, high quality
gpt-image-1	1024x1024, 512x512, 256x256	Supports generation and editing

Gemini / Google

Auto-discovers which image models are available on your API key:

Model	Notes
gemini-2.5-flash-image	Fast generation
gemini-3-pro-image-preview	Higher quality
imagen-4.0-generate-001	Dedicated image model
imagen-4.0-ultra-generate-001	Highest quality
imagen-4.0-fast-generate-001	Fastest Imagen

---

How It Works

OpenAI — POSTs to /v1/images/generations or /v1/images/edits, receives base64-encoded image data, decodes and saves as PNG.

Gemini — uses a multi-strategy approach:

1. Tries Imagen models via predict and generateImages endpoints
2. Tries Gemini Flash via generateContent with responseModalities: ["IMAGE"]
3. Calls ListModels API to auto-discover all image-capable models on your key
4. On HTTP 429, parses retry delay and waits automatically (up to 3 retries)

Image editing — OpenAI uses multipart form upload to /v1/images/edits. Gemini sends the source image as base64 inline data via generateContent or Imagen's predict endpoint.

---

Environment Variables

Variable	Description
OPENAI_API_KEY	OpenAI API key (starts with sk-)
GEMINI_API_KEY	Google Gemini API key (starts with AIza)

At least one must be set. If both are set, the script lets you choose the provider per generation.

---

Troubleshooting

Issue	Cause	Fix
No API keys found	Keys not set as env vars	Set OPENAI_API_KEY or GEMINI_API_KEY before running
Gemini rate limited / 429	Free tier — quota is 0 for image gen	Enable billing at ai.google.dev
Gemini 404 for all models	Region restriction or key not propagated	Wait a few minutes and retry, or use OpenAI
OpenAI insufficient quota	Account credits exhausted	Add funds at platform.openai.com/account/billing
Unsupported image format	File is not PNG, JPG, JPEG, or WebP	Convert the image before editing
Request timed out (120s)	Slow generation or connection issue	Check connection, try a simpler prompt, or switch provider

---

License

MIT