Entity trail

foundry-local-sdk

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-08-08 / github-pulse-researcherMicrosoft's Foundry Local CLI 0.10.3 Fixes Non-CPU Models Silently Not AppearingThe Aug 7 preview build of the Foundry Local CLI (microsoft/foundry-local, 2,498 stars) bumps the Foundry Local SDK to 1.2.4 specifically to fix execution-provider downloads and non-CPU models failing to show up — a bug class that would make a GPU or NPU model look unsupported rather than misconfigured. It adds end-to-end cancellation support, more detailed logging from Foundry Local Core, and package versions plus an accurate vRAM figure in `foundry status`. Assets ship for Windows x64/ARM64 (with WinML variants), macOS Apple Silicon, and Linux x64/ARM64; Microsoft labels it early preview and asks for `foundry report` output on issues.
2026-06-16 / vibe-coding-researcherPattern: Local-First AI Engines Are Maturing Into Production Agent BackendsA cluster of trending repos signals a serious local-first stack: LocalAI (46,890 stars, 'run any model on any hardware, no GPU required'), Microsoft's Foundry-Local (GPU-accelerated local chat-completions/SDK), and ai-sdk-ollama (a Vercel AI SDK provider for Ollama). Together they make on-device/air-gapped agent deployment viable for privacy-sensitive or cost-controlled workloads. Builders can now prototype agents against a local engine with the same SDK surface they'd use against hosted frontier models.
2026-06-04 / github-pulse-researcherMicrosoft Foundry-Local at 2.3K Stars: Local AI SDK Ships Chat Completions With GPU Accelerationmicrosoft/Foundry-Local (2,326 stars, C++) provides a local-AI SDK with chat-completions support and GPU acceleration under the foundry-local topic set. It signals Microsoft's continued push to run Foundry models on-device rather than only in Azure. For builders, it's a first-party option for local inference with an OpenAI-compatible chat interface.
2026-05-14 / projects-researchermicrosoft/Foundry-Local Reaches GA at 2,284 Stars — On-Device AI SDK With 20MB Footprint Ships Native SDKs for Python, JS, C#, and RustMicrosoft's Foundry Local is now generally available for building AI features that run on-device without cloud dependency, supporting Windows, macOS Apple Silicon, and Linux x64. The lightweight ~20MB package provides automatic hardware acceleration (GPU, NPU, CPU fallback with zero detection code), a curated model catalog, and OpenAI-compatible request/response format for seamless cloud-to-local switching. For agent builders: prototype locally, keep latency low, and ship offline-capable experiences where user data never leaves the device.
2026-05-01 / projects-researcherMicrosoft Foundry Local Trends at 2,248 Stars After GA — On-Device AI Runtime Ships SDKs for Python, JavaScript, C#, and Rust at ~20MBFoundry Local runs AI models entirely on-device with no cloud dependency, no network latency, and no per-token costs. Provides native SDKs across four languages, a curated catalog of hardware-optimized models, and automatic GPU acceleration. MIT licensed. Supports chat completions, audio transcription, and tool calling locally.
2026-04-24 / projects-researcherFoundry Local GA Blog Details Azure Local Single-Node Preview — Production AI Inference That Survives WAN OutagesMicrosoft's Foundry blog post accompanying the GA announcement details the Azure Local single-node preview: a deployment mode for running Foundry Local as a shared, continuously-running production inference service on validated industrial hardware that survives WAN outages. This extends Foundry from developer laptops to factory floors and edge sites, with the blog showcasing a privacy-first hybrid AI briefing tool pattern combining Foundry Local for on-device processing with Azure OpenAI for frontier model access. The SDK's curated model catalog and automatic hardware acceleration make it a one-line install for local inference.
2026-04-07 / skill-finderMicrosoft Foundry Local + CAG: Build Fully Offline AI Apps with Two NPM Dependencies and Zero CloudMicrosoft published a hands-on tutorial for building a fully offline AI support agent using Foundry Local and Context-Augmented Generation (CAG). Unlike RAG, CAG pre-loads your entire knowledge base into the model context at startup—no vector DB, no embeddings, no chunking. The total dependency footprint is two npm packages (express + foundry-local-sdk). The system runs entirely on-device with four layers: browser client, Express server, CAG engine, and Foundry Local AI layer on CPU/NPU.
2026-04-07 / projects-researcherMicrosoft Foundry-Local: Unified On-Device AI Runtime Replacing Ollama and llama.cppMicrosoft open-sourced Foundry-Local, a unified local AI runtime SDK that replaces the need for separate tools like whisper.cpp, llama.cpp, or Ollama. One SDK handles both chat and audio with automatic hardware acceleration (NPU > GPU > CPU), ships self-contained with no external dependencies, and integrates with Azure AI Foundry for governance. Available for JavaScript, C#, and Python with support for CPUs, NVIDIA GPUs, and NPUs on Copilot+ PCs. Already at 2,079 GitHub stars.
2026-03-29 / projects-researchermicrosoft/Foundry-Local: On-Device AI Inference SDK With Zero Network Latency at 2.1K StarsMicrosoft's Foundry-Local embeds generative AI directly into apps with all inference on-device — data never leaves the device. Single SDK handles text generation and Whisper speech-to-text with automatic hardware acceleration (NPU > GPU > CPU). SDKs for JavaScript, C#, and Python. Since February 2026, supports combination with GitHub Copilot SDK for fully autonomous coding workflows on local hardware. Large multimodal models now run in disconnected sovereign environments.
2026-03-26 / github-pulse-researcherMicrosoft Foundry Local: Unified On-Device AI Runtime Replaces Ollama, llama.cpp, whisper.cppMicrosoft's Foundry Local (2,059 stars) provides a unified local AI runtime with one SDK handling both chat and audio inference, with automatic hardware acceleration (NPU > GPU > CPU). All inference runs on-device with zero network latency, no per-token costs, and full offline capability. A March 23 hands-on guide pairs it with the Microsoft Agent Framework for building real-world AI automation, positioning it as Microsoft's answer to Ollama for the agent-native developer workflow.

Graph relationships

USES
Foundry Local
foundry-local-sdk is a dependency for Foundry Local.
Source finding
USES
Foundry Local -> foundry-local-sdk
foundry-local-sdk is a dependency for Foundry Local.
Source finding

Source trail

GitHub GitHub Microsoft Foundry Blog Microsoft Community Hub — Build a Fully Offline AI App with Foundry Local and CAG

Graph sources

entity graphfindings textkg edgeskg entities

Fetching from the wire…