Installation

Installation

Get started with WebLLM.io by installing the core SDK package.

Core Package

Install @webllm-io/sdk using your preferred package manager:

# npm
npm install @webllm-io/sdk

# pnpm
pnpm add @webllm-io/sdk

# yarn
yarn add @webllm-io/sdk

Optional Peer Dependency

For local inference support, you need to install @mlc-ai/web-llm as a peer dependency:

# npm
npm install @mlc-ai/web-llm

# pnpm
pnpm add @mlc-ai/web-llm

# yarn
yarn add @mlc-ai/web-llm

Note

If you only plan to use cloud inference, you can skip installing @mlc-ai/web-llm. The SDK will automatically disable local inference when the peer dependency is not available.

TypeScript Support

TypeScript definitions are included out of the box. No additional @types packages are needed.

Browser Requirements

Local Inference (WebGPU)

To run models locally in the browser, you need:

• Chrome 113+ or Edge 113+
• WebGPU support enabled (usually on by default)
• Sufficient VRAM — Device scoring adapts model selection to available resources

Check Device Capability

Use the built-in capability checker to verify WebGPU support:

import { checkCapability } from '@webllm-io/sdk';

const capability = await checkCapability();
console.log(capability.grade); // S, A, B, or C
console.log(capability.canRunLocal); // true if WebGPU is available

Cloud Mode

Cloud inference works in all modern browsers with no special requirements:

• Chrome, Firefox, Safari, Edge (latest versions)
• Mobile browsers (iOS Safari, Chrome Mobile)

Verification

After installation, verify everything is set up correctly:

import { createClient } from '@webllm-io/sdk';

const client = createClient({
  local: 'auto',
  cloud: { baseURL: 'https://api.openai.com/v1', apiKey: 'sk-...' },
});

console.log('WebLLM client created successfully!');

Next Steps

• Quick Start — Make your first completion
• Playground — Try the interactive demo
• Configuration — Learn about advanced options