VS Code Language Model API

Sypha provides experimental support for the VS Code Language Model API, a standardized interface that allows different extensions to share model access. This bridge enables you to utilise models authorized by other services already installed in your editor, such as:

• GitHub Copilot: Utilising your existing Copilot subscription.
• Specialized Extension Suites: Any third-party tool that implements the VS Code LM registry.

[!IMPORTANT] This integration is currently in an experimental state. Reliability is directly dependent on how well external extensions implement the VS Code standard.

Pre-Configuration Requirements

1. Modern VS Code: Ensure you are running a recent version of VS Code.
2. Active Provider Extension: You must have a supporting extension installed and authenticated (e.g., GitHub Copilot and its Chat component).

Configuring Sypha

1. Activate Settings: Click the gear icon in the Sypha sidebar.
2. Set Provider: Choose VS Code LM API from the provider registry.
3. Assign Model: Browse the resulting list of available models. These are typically formatted as Vendor / Model Family (e.g., copilot - claude-3.5-sonnet).

Strategic Operational Factors

• Extensibility: Sypha cannot directly control this model list; it only reflects what other extensions have authorized for sharing.
• Capability Nuance: The VS Code LM API may not support advanced features like high-fidelity image processing or ultra-long streaming as robustly as direct API providers.
• Fiscal Transparency: Interaction costs and quotas are governed by the extension providing the model. Sypha does not track individual token expenditure for these shared sessions.
• Rate Limit Sync: If using GitHub Copilot, you will still be subject to GitHub’s aggregate rate limits.

Common Resolution Steps

• Registry Latency: If no models appear, ensure the provider extension (like Copilot) is fully authenticated. In some cases, sending a single message via the provider's native chat interface initializes the API for Sypha.
• Unexpected Interaction: If a shared model behaves inconsistently, it often indicates a mismatch between the editor's LM API version and the provider's implementation.

---

Visit the Official VS Code LM Documentation for technical updates on the standard.