GitHub Copilot with ARK

GitHub Copilot: The AI chat and agent in Visual Studio Code, backed by GitHub Copilot.

Before you start

You need an ARK API key: sign in to ARK Cloud and copy it from the dashboard, or use a key from your own ARK deployment and its base URL. The examples use arklabsai/Qwen3.6-35B-A3B; for the hardest tasks use arklabsai/Qwen3.5-122B-A10B. Both call tools and read up to 256K tokens.

1. Install

You need Visual Studio Code 1.122 or later, where the Custom Endpoint provider is stable. Models you add yourself work without a Copilot plan or a GitHub sign-in.

2. Connect it to ARK

Run Chat: Manage Language Models from the Command Palette, choose Add Models and then Custom Endpoint. Enter a group name, a display name and your ARK key, and pick Chat Completions as the API type. VS Code opens chatLanguageModels.json; make the ARK entry look like this:

[
  {
    "name": "ARK",
    "vendor": "customendpoint",
    "apiKey": "${input:arkApiKey}",
    "apiType": "chat-completions",
    "models": [
      {
        "id": "arklabsai/Qwen3.6-35B-A3B",
        "name": "Qwen3.6 35B A3B (ARK)",
        "url": "https://api.ark-labs.cloud/api/v1/chat/completions",
        "toolCalling": true,
        "contextWindow": 262144,
        "maxOutputTokens": 8192
      }
    ]
  }
]

Keep the apiKey reference VS Code wrote for the key you entered. Save the file and pick Qwen3.6 35B A3B (ARK) in the chat model picker; if it does not appear, restart VS Code.

Good to know

• Give the full chat completions URL: VS Code uses a URL with an explicit API path as it is.
• toolCalling: true makes the model available to agents: VS Code offers only tool-calling models there.
• Inline suggestions, semantic search and features that rely on embeddings still run on GitHub's models and need a GitHub account.
• On Business and Enterprise plans, the Bring Your Own Language Model Key in VS Code policy governs Custom Endpoint models; ask your administrator if the option is missing.
• Code completion while you type needs a fill-in-the-middle endpoint, which ARK does not serve: use ARK for chat and agent work.
• Agents offer tools on every request, and while tools are offered the Qwen models answer without thinking (see Reasoning).
• These tools send the whole conversation on every turn, so they use ARK statelessly. ARK's prompt cache speeds up the repeated part of each prompt; every prompt token is still billed.

Next

Read about tool calling and limits and capacity on ARK, or browse other integrations.

Configuration keys follow GitHub Copilot's official documentation as of October 2026. If the tool has changed since, its docs take precedence.