Stability AI Provider for Vercel AI SDK

A custom Vercel AI SDK provider for Stability AI image generation models.

Installation

npm install stability-ai-provider ai
# or
bun add stability-ai-provider ai
# or
pnpm add stability-ai-provider ai

Setup

Set your Stability AI API key as an environment variable:

export STABILITYAI_API_KEY="your-api-key"

Or pass it directly when creating the provider:

import { createStabilityAI } from "stability-ai-provider";

const stabilityai = createStabilityAI({
  apiKey: "your-api-key",
});

Usage

Basic Image Generation

import { stabilityai } from "stability-ai-provider";
import { generateImage } from "ai";

const { image } = await generateImage({
  model: stabilityai.image("sd3.5-large"),
  prompt: "A serene mountain landscape at sunset with vibrant colors",
});

console.log(image.base64); // Base64-encoded image data

With Options

import { stabilityai } from "stability-ai-provider";
import { generateImage } from "ai";

const { images } = await generateImage({
  model: stabilityai.image("sd3.5-large"),
  prompt: "A cyberpunk city at night",
  n: 2, // Generate 2 images
  aspectRatio: "16:9",
  seed: 12345, // For reproducible results
  providerOptions: {
    "stability-ai": {
      stylePreset: "neon-punk",
      cfgScale: 8,
      steps: 50,
      negativePrompt: "blurry, low quality",
    },
  },
});

Custom Provider Instance

import { createStabilityAI } from "stability-ai-provider";
import { generateImage } from "ai";

const stabilityai = createStabilityAI({
  apiKey: process.env.MY_STABILITY_KEY,
  headers: {
    "X-Custom-Header": "value",
  },
});

const { image } = await generateImage({
  model: stabilityai.image("core"),
  prompt: "A beautiful forest",
});

Supported Models

| Model ID | Description | |----------|-------------| | sd3.5-large | Stable Diffusion 3.5 Large - Highest quality | | sd3.5-large-turbo | Stable Diffusion 3.5 Large Turbo - Fast generation | | sd3.5-medium | Stable Diffusion 3.5 Medium - Balanced | | sd3-large | Stable Diffusion 3 Large | | sd3-large-turbo | Stable Diffusion 3 Large Turbo | | sd3-medium | Stable Diffusion 3 Medium | | core | Stable Image Core - General purpose |

Provider Options

Pass these in providerOptions["stability-ai"]:

| Option | Type | Default | Description | |--------|------|---------|-------------| | stylePreset | string | - | Style preset (see below) | | cfgScale | number | 7 | Prompt adherence (1-35) | | steps | number | 50 | Inference steps (10-150) | | outputFormat | string | "png" | Output format: "png", "jpeg", "webp" | | negativePrompt | string | - | What to avoid in generation |

Style Presets

Available style presets:

• enhance - General enhancement
• anime - Anime style
• photographic - Photorealistic
• digital-art - Digital art style
• comic-book - Comic book style
• fantasy-art - Fantasy art
• line-art - Line art
• analog-film - Film photography look
• neon-punk - Neon/cyberpunk style
• isometric - Isometric view
• low-poly - Low polygon style
• origami - Paper craft style
• modeling-compound - Clay/modeling look
• cinematic - Cinematic style
• 3d-model - 3D rendered look
• pixel-art - Pixel art style
• tile-texture - Seamless texture

API Reference

createStabilityAI(options?)

Creates a new Stability AI provider instance.

Options:

• apiKey?: string - API key (defaults to STABILITYAI_API_KEY env var)
• baseURL?: string - Custom API base URL
• headers?: Record<string, string> - Custom headers
• fetch?: typeof fetch - Custom fetch implementation

stabilityai.image(modelId, settings?)

Creates an image model instance.

Parameters:

• modelId: string - The model ID to use
• settings?: StabilityAIImageSettings - Default settings for this model

Requirements

• Node.js 18+
• Stability AI API key
• ai package (peer dependency)

License

MIT