Sign in from the browser and store the API key securely in VS Code SecretStorage.
Use OpenProvider models from the Copilot Chat model picker.
Route through openprovider/auto-free or select exact working models from OpenProvider.
Track token usage and estimate costs in real-time.
Supports streaming, tool calls, image-capable models, and reasoning output when the selected model exposes those capabilities.
Requirements
VS Code 1.120.0 or newer.
GitHub Copilot Chat enabled, because VS Code language model providers surface through the Copilot Chat model picker.
Login flow
Run OpenProvider: Sign in to OpenProvider. The extension opens:
https://openprovider.mimika.in/api/vscode/connect
After Google sign-in, OpenProvider creates a VS Code API key and posts it back
to a temporary localhost callback server. The key is stored in VS Code
SecretStorage, so users do not need to copy or configure an API key manually.
Development
npm install
npm run compile
npm run esbuild
Request settings
openprovider.outputTokenBudget controls the normal max_tokens request budget. The extension clamps it to the selected model limit.
openprovider.retryOutputTokenBudget controls compact retry requests after upstream rejections or empty streams.
openprovider.forwardSamplingParameters is disabled by default so routed providers that reject temperature or top_p keep working. Enable it only when the selected provider supports those parameters.
