Integrations

GitHub Copilot

GitHub Copilot's chat panel does not natively accept a custom API endpoint. To route Copilot Chat through ezrouter, install a community VS Code extension that registers extra models in the Copilot model picker and forwards their calls to a custom base URL.

Prerequisites

  • VS Code 1.116 or newer.
  • A GitHub Copilot subscription (Free, Pro, or Enterprise).
  • An ezrouter API key from the

dashboard.

Install a bridge extension

Install a Copilot-Chat bridge that supports a custom base_url. The community-maintained deepseek-v4-for-copilot extension is one option — it adds models to the Copilot picker and lets you point them at any OpenAI-compatible endpoint, including ezrouter.

After installing, open the Command Palette (Cmd+Shift+P / Ctrl+Shift+P) and look for the extension's configuration commands. Most bridge extensions expose two settings:

  • A base URL field — set to https://www.ezrouter.dev/v1.
  • An API key field — paste your ezrouter key. The extension stores

it in the OS keychain.

Use a model

Open Copilot Chat (Cmd+Shift+I / Ctrl+Shift+I), click the model picker at the top of the chat panel, and select the bridge-registered model. Copilot's agent mode, tool calling, and MCP features run on top of the chosen model.

Other approaches

If you do not want a third-party extension, two other paths exist:

  • Switch your global Copilot model in

VS Code settings if your Copilot tier exposes a "Bring your own model" preview. This is a GitHub-side feature and is not always available.

OpenCode accept a base URL directly without an intermediate extension.

Troubleshooting

  • Model does not appear in the picker — confirm the extension is

enabled and that VS Code was reloaded after installing it.

  • 401 invalid_api_key — re-enter the key in the extension's

command. Keys with trailing whitespace silently fail.

  • 403 access_denied — the model the extension is requesting is

not in your key's allowlist. Check GET /v1/models.

Next steps

the bridge extension uses.

  • Pricing — per-model rates.