GitHub Copilot BYOK (VS Code) with a local model: setup for Ollama, LM Studio or llama.cpp

Bring Your Own Key (BYOK) in VS Code lets the built-in chat and agent use models you supply, including local ones, without a Copilot plan or GitHub sign-in. Ollama is connected through the official Ollama extension, and other local servers through the Custom Endpoint provider. It suits VS Code users who want to keep the native chat UI; inline suggestions, semantic search and embeddings still need GitHub's service.

Official docsCompare all agents

At a glance

Type
vscode-chat
Licence
Not verified
Latest release
Not verified
Status
Maintained
Local backends
ollama, openai-compatible, anthropic-compatible
Needs native tool calling
Yes
MCP support
Not verified
Recommended minimum context
Not stated in its docs
Start-up prompt
Not measured
GitHub stars
Not verified

How to set it up with a local model

Pick your backend. Each command has its own Copy button. Where a step uses an example model, swap in the one you run; the table below lists the ones that fit your machine.

  1. Install it (Ollama extension (VS Code CLI)):

    code --install-extension Ollama.ollama

  2. Install the official Ollama extension. The built-in Ollama provider in VS Code is deprecated.

    code --install-extension Ollama.ollama

  3. Pull a model that supports tool calling (replace qwen3-coder:30b with your choice).

    ollama pull qwen3-coder:30b

  4. Raise Ollama's context window (the Ollama VS Code guide asks for at least 64K); or set the context slider in the Ollama app's Settings instead.

    OLLAMA_CONTEXT_LENGTH=65536 ollama serve

  5. Reload the VS Code window, open Chat and choose the model from the Ollama section of the model picker.

Check that it works

Open Chat in agent mode, pick the local model and ask it to read a file. If an Ollama model is missing, run Ollama: Refresh Models and Ollama: Diagnose Models from the Command Palette.

ollama list

Which local models fit, by machine

Checked by this site's engine at the context GitHub Copilot BYOK needs, with 32 GB of system RAM for MoE offload. A Mac counts three quarters of unified memory as usable.

MachineModelQuantContextFitOllama
12 GB GPUQwen 3.5 9BQ4_K_M32,768Comfortable (7.4 GB of 12 GB)qwen3.5:9b
Qwen 3.6 35B-A3BQ4_K_M32,768With CPU offload (MoE) (22.7 GB of 12 GB)qwen3.6:35b-a3b
16 GB GPUQwen 3.5 9BQ4_K_M32,768Comfortable (7.4 GB of 16 GB)qwen3.5:9b
Qwen 3.6 35B-A3BQ4_K_M32,768With CPU offload (MoE) (22.7 GB of 16 GB)qwen3.6:35b-a3b
24 GB GPUQwen3.8 27BQ4_K_M32,768Comfortable (19.9 GB of 24 GB)qwen3.8:27b
Qwen 3.6 27BQ4_K_M32,768Comfortable (19.9 GB of 24 GB)qwen3.6:27b
Devstral Small 2 24BQ4_K_M32,768Comfortable (20.7 GB of 24 GB)devstral-small-2:24b
Qwen 3.6 35B-A3BQ4_K_M65,536Tight (23.3 GB of 24 GB)qwen3.6:35b-a3b
32 GB GPUQwen3.8 27BQ4_K_M32,768Comfortable (19.9 GB of 32 GB)qwen3.8:27b
Qwen 3.6 27BQ4_K_M32,768Comfortable (19.9 GB of 32 GB)qwen3.6:27b
Devstral Small 2 24BQ4_K_M32,768Comfortable (20.7 GB of 32 GB)devstral-small-2:24b
Qwen 3.6 35B-A3BQ4_K_M65,536Comfortable (23.3 GB of 32 GB)qwen3.6:35b-a3b
Qwen3-Coder 30B-A3B (MoE)Q4_K_M65,536Comfortable (25.7 GB of 32 GB)qwen3-coder:30b
Mac, 64 GBQwen 3.6 35B-A3BQ8_0131,072Comfortable (40.7 GB of 48 GB)—

Your exact machine: filter the matrix or check any model in the analyzer.

Common problems with local models

  • Agent mode only lists models that support tool calling; for a Custom Endpoint model, set "toolCalling": true.
  • VS Code may show a model's maximum context while Ollama allocates less at runtime; raise Ollama's context (64K or more) and reload the window.
  • Title generation, commit messages and other utility features keep using GitHub's models unless you point chat.utilityModel and chat.utilitySmallModel at a local model.
  • Copilot Business and Enterprise users need an admin to enable the Bring Your Own Language Model Key policy, and local models cannot drive inline suggestions.

Other coding agents

How they rank for local models: the best agent harness for local LLMs.

Sources (9), checked 2026-10-20