LM Studio AI SDK Provider

LM Studio provider for the Vercel AI SDK with proper token usage tracking.

Why This Provider?

The OpenAI-compatible client doesn't track token usage during streaming. This provider uses the LM Studio TypeScript SDK which has better support and proper token tracking.

Requirements

• Node.js 18+
• LM Studio running locally or remotely

Installation

npm install @djosh34/lmstudio-ai-sdk-provider

Usage

import { createLMStudio } from '@djosh34/lmstudio-ai-sdk-provider';
import { generateText, streamText } from 'ai';

// Create provider (uses localhost:1234 by default)
const lmstudio = createLMStudio();

// Generate text
const { text } = await generateText({
  model: lmstudio('your-model-name'),
  prompt: 'Hello, world!',
});

// Stream text
const { textStream } = await streamText({
  model: lmstudio('your-model-name'),
  prompt: 'Tell me a story',
});

for await (const delta of textStream) {
  process.stdout.write(delta);
}

Tool Calling Example

import { createLMStudio } from '@djosh34/lmstudio-ai-sdk-provider';
import { generateText, tool } from 'ai';
import { z } from 'zod';

const lmstudio = createLMStudio();

const { text } = await generateText({
  model: lmstudio('your-model-name'),
  prompt: 'Calculate 15 + 27',
  tools: {
    calculator: tool({
      description: 'A calculator tool',
      parameters: z.object({
        a: z.number(),
        b: z.number(),
      }),
      execute: async ({ a, b }) => a + b,
    }),
  },
});

Configuration

import { createLMStudio } from '@djosh34/lmstudio-ai-sdk-provider';

// Custom LM Studio server
const lmstudio = createLMStudio({
  // LM Studio connection settings go here
  // See LM Studio TypeScript SDK docs for options
});

// Provider-specific options
const { text } = await generateText({
  model: lmstudio('your-model-name'),
  prompt: 'Hello!',
  providerOptions: {
    lmstudio: {
      contextOverflowPolicy: 'stopAtLimit',
      // Other LM Studio specific options
    }
  }
});

What Works

• Text generation
• Text streaming
• Tool calling
• Token usage tracking
• Reasoning parsing (for supported models)
• All LM Studio-specific features

What Doesn't Work

• Image generation (LM Studio limitation)
• Embeddings (use dedicated embedding models)

API Reference

createLMStudio(settings?)

Creates an LM Studio provider instance.

• settings: Optional LM Studio connection and model settings

Provider Methods

• lmstudio(modelId, settings?) - Create a chat model
• lmstudio.chat(modelId, settings?) - Alias for the above
• lmstudio.languageModel(modelId, settings?) - Alias for the above

License

Apache-2.0