OVHcloud AI Endpoints for Copilot Chat

Use the models hosted on OVHcloud AI Endpoints (Llama, Mistral, Qwen, DeepSeek, gpt-oss, ...) directly inside GitHub Copilot Chat in VS Code and GitHub Codespaces. European, sovereign AI inference for your editor.

Published by EUSovereignAI. This is a community extension, not affiliated with or endorsed by OVHcloud. It is the VS Code companion of OVHcloudAIEndpointsTester.

Features

• Adds an OVHcloud AI Endpoints provider to the Copilot Chat model picker.
• Models are discovered automatically from the /v1/models endpoint.
• Streaming responses, tool calling (agent mode) and image input for vision models.
• Your API token is stored in VS Code secret storage, never in settings files.
• Works with any OpenAI-compatible router by changing ovhAiEndpoints.baseUrl.

Setup

An interactive version of this guide is available in VS Code: Command Palette > OVHcloud AI Endpoints: Open setup guide (or Help > Get Started > Set up OVHcloud AI Endpoints for Copilot Chat). The extension also shows a notification the first time it runs without a token.

1. Get an OVHcloud AI Endpoints token

1. Open the OVHcloud control panel.
2. Select your Public Cloud project.
3. In the left menu, go to AI & Machine Learning > AI Endpoints, then the API keys tab.
4. Create a key and copy it. Anonymous access exists but is limited to 2 requests/minute and is not supported here.

2. Store the token in VS Code

The token is not a setting. It lives in VS Code secret storage (macOS Keychain, Windows Credential Manager, libsecret on Linux), so it never ends up in settings.json, in your dotfiles, or in a git repository.

Any of these three ways works:

• Command Palette (Cmd+Shift+P / Ctrl+Shift+P) > OVHcloud AI Endpoints: Set API token, paste the key, press Enter.
• Copilot Chat > click the model name > Manage Models... > OVHcloud AI Endpoints. If no token is stored, the same input box opens.
• Command Palette > OVHcloud AI Endpoints: Manage API token for a small menu (set, test, clear, settings, guide).

To replace the token, run Set API token again. To remove it, run Clear API token.

3. Check the connection

Command Palette > OVHcloud AI Endpoints: Test connection and list models. You should see the list of models your token gives access to. Errors (wrong token, rate limit, wrong base URL) are shown in a notification and logged in the OVHcloud AI Endpoints output channel (View > Output).

4. Pick a model in Copilot Chat

Copilot Chat > click the model name > Manage Models... > OVHcloud AI Endpoints, tick the models you want. They now appear in the model picker like any other model, with streaming, tool calling (agent mode) and image input for vision models.

Settings

Commands

• OVHcloud AI Endpoints: Set API token
• OVHcloud AI Endpoints: Clear API token
• OVHcloud AI Endpoints: Test connection and list models

Requirements

• VS Code 1.104 or newer.
• GitHub Copilot Chat extension.
• An OVHcloud AI Endpoints access token (anonymous access is rate limited and not supported here).

License

MIT