Skip to main content
Foundry Local logo Foundry Local logo

Foundry Local

Build once, run locally. The SDK for shipping AI-powered applications with hardware-optimized on-device inference.

Get Started in Two Steps

Step 1: Install Foundry Local
brew install microsoft/foundrylocal/foundrylocal
Step 2: Run a model
foundry model run qwen2.5-0.5b
Docs Models

Purpose built for shipping AI applications

Everything you need to embed AI into your products, with the performance and reliability your users expect.

Ship to Production

Built as an SDK for shipping AI-powered applications, not just running models locally

// Initialize & Load
const manager = FoundryLocalManager.create(config)
const model = manager.getCatalog().getModel('gpt-oss-20b')
await model.load()|

Hardware Optimized

We work directly with hardware vendors for maximum performance

NPU
Neural Engine
GPU
Graphics Card
CPU
Processor

Edge-Ready

Works fully offline with no cloud dependencies

Multi-Language SDKs

Python, JavaScript, C#, and Rust

PY
JS
C#
RS

OpenAI Compatible

Drop-in API replacement for easy integration

base_url="api.openai.com"
base_url="localhost"

Data Privacy

Everything stays on-device