Foundry Local runs Qwen3.8-27B on your own GPU and plugs it into GitHub Copilot
What is it?
Microsoft published a setup for running Qwen3.8-27B locally with Foundry Local, its on-device inference tool, and connecting it to GitHub Copilot through BYOK (bring your own key). It runs on ONNX Runtime and uses speculative decoding, where a small drafter model guesses several tokens ahead, plus KV-cache quantization to keep it fast and within memory. You need Foundry Local CLI 0.13.0 or later. Install it, load the model, then pick Foundry Local under Model providers in the Copilot app.
winget install Microsoft.FoundryLocal
foundry --version
foundry model load qwen3.8-27b
foundry server statusWhy does it matter?
At 4-bit the model shrinks to about 18 GBVendor claim, which Microsoft says fits a 24 GB GPU with up to 64K context, or the full 262K on 32 GB and up. Microsoft lists Qwen3.8-27B at 73.0 on Terminal Bench 2.1Vendor claim and 61.7 on SWE-bench ProVendor claim. Teams that can't send code to the cloud, or developers who work offline, get a coding agent with no per-token bill.
Maturity
PreviewNVIDIA GPUs with 24 GB or more only, and you install CUDA and cuDNN yourself. Mac and AMD support is still on the roadmap, so we're calling this a preview. Performance numbers are Microsoft's own.
How to track it
- Hardware: watch the Foundry Local release notes for Mac and AMD GPU support.
- Copilot: see whether GitHub's BYOK docs add Foundry Local as a built-in provider.
- Real-world speed: look for developers posting their own Foundry Local vs llama.cpp measurements.