# microsoft/foundry-local

Repository: https://github.com/microsoft/foundry-local
Canonical: https://ross.abutalabs.com/products/foundry-local
Homepage: https://foundrylocal.ai
Language: C++
License: NOASSERTION
License Family: other
Topics: ai-sdk, chat-completions, foundry-local, gpu-acceleration, local-ai, microsoft, on-device-inference, onnx-runtime, speech-to-text, whisper
Last push: 2026-08-26T21:28:23+00:00

## Health v2 (maintenance only)
Score: 85/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 99, release rhythm 95, longevity 37
- inputs: {"age_days": 520, "days_push": 7, "days_rel": 34, "gap_med": 24.0, "n_releases_24m": 5}
- flags: no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2525, forks 362 (observed 2026-08-28T04:06:58.512676+00:00)

## What it is
Foundry Local is Microsoft's end-to-end local AI runtime and SDK suite (C#, JavaScript, Python, Rust) for running optimized models entirely on-device via ONNX Runtime. It bundles a curated model catalog, automatic NPU/GPU/CPU hardware acceleration, model lifecycle management, and an optional OpenAI-compatible local server.

## Use cases
- run LLM chat completions locally without an API key or cloud backend
- transcribe audio on-device with Whisper
- ship offline AI features inside a desktop or mobile app
- point an existing OpenAI SDK app at a local endpoint
- serve local models to LangChain via an OpenAI-compatible server
- automatically pick the best execution provider for NPU, GPU, or CPU

## When to choose
- you need on-device inference with data never leaving the user's machine
- you want zero network latency, offline operation, and no per-token costs
- you want automatic hardware acceleration and model management with minimal code
- your app already uses the OpenAI SDK and you want a drop-in local endpoint

## When to avoid
- you need frontier-scale models that exceed consumer device hardware
- you require a permissive open-source license (license is custom/unspecified)
- you need multi-node or cloud-scale serving rather than single-device inference

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, speech-recognition, sdk, http-server, machine-learning
- domain: large-language-models, artificial-intelligence, speech-processing, developer-tools
- platform: windows, cross-platform, python, rust, dotnet
- tags: on-device-ai, onnx-runtime, openai-compatible, local-inference, model-catalog, hardware-acceleration, offline-ai, whisper, macos, nodejs, gpu

## Member repositories
- microsoft/foundry-local (main) score 85

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:58.512676+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T02:25:32.306423+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/foundry-local (fetched 2026-08-28T04:06:58.512676+00:00, sha 2911db2ccc15)
  - homepage: https://foundrylocal.ai (fetched 2026-08-29T10:07:57.054803+00:00, sha cdd6555e1a95)
- Data as of 2026-08-30T08:39:29.467469+00:00.