Provider Protocol

Multi-Provider

One interface, any provider, any surface. Your backend code is your frontend code - your OpenAI code is your anthropic code.

Type-Safe

Full TypeScript support with intelligent autocompletion and compile-time checks.

Transparent

Build without worrying about magic, read provider official documentation and you already know how to use Provider Protocol.

Modular

Purpose-built interfaces for llm(), embedding(), and image().

Quick Start

import { llm, embedding } from '@providerprotocol/ai';
import { anthropic } from '@providerprotocol/ai/anthropic';
import { openai } from '@providerprotocol/ai/openai';

// LLM inference with Claude
const claude = llm({
  model: anthropic('claude-sonnet-4-20250514'),
  system: 'You are a helpful assistant.',
});

const response = await claude.generate('Explain quantum computing');
console.log(response.text);

// Generate embeddings with OpenAI
const embedder = embedding({
  model: openai('text-embedding-3-large'),
});

const vector = await embedder.embed('Hello, world!');

Specification

UPP-1.3 Current

Interface	Status	Description
LLM	Stable	Text generation, streaming, tools, structured output
Embedding	Stable	Vector embeddings, batch processing, similarity
Image	Stable	Generation, editing, variations
Audio	Planned	Speech-to-text, text-to-speech

---

Introduction

Learn about Provider Protocol’s goals and architecture.

Quick Start

Follow the Quick Start guide to begin building.