Bring DevPass / LLM Gateway models into VS Code's built-in Chat with just an API key.
DevPass registers itself as a language model provider in VS Code, so the models you have access to through LLM Gateway show up directly in the Chat model picker. The endpoint is pre-configured (https://api.llmgateway.io/v1) and your available models are discovered automatically — there's nothing to wire up beyond pasting your key.
Features
- Automatic model discovery — your available models appear in the Chat model picker; deprecated and deactivated models are filtered out.
- Streaming responses — answers stream into Chat as they're generated.
- Tool / function calling — supported per model, so agentic workflows and tools work out of the box where the model allows it.
- Image input — vision-capable models accept images.
- Secure key storage — your API key is kept in VS Code's encrypted SecretStorage, never in settings or plaintext.
- Resilient networking — transient failures are retried with backoff, honoring the server's
Retry-Afterhint, with timeouts and stall detection on streams. - Zero endpoint config — no base URLs or model lists to maintain; just add your key.
Getting started
- Install the extension.
- Run DevPass: Set API Key from the Command Palette and paste your key (format:
llmgtwy_...). Alternatively, just pick a DevPass model in Chat and you'll be prompted for the key the first time. - Open the Chat view and select a DevPass model from the model picker.
- Start chatting.
Commands
- DevPass: Manage API Key — set/update your key or sign out from a single menu.
- DevPass: Set API Key — enter or replace your stored key.
- DevPass: Sign Out (Clear API Key) — remove the stored key.
Requirements
- VS Code
1.104.0or newer. - A DevPass / LLM Gateway API key from your DevPass / LLM Gateway account.
Troubleshooting
If models don't load or responses fail, open View → Output and select the DevPass channel. It logs retries, cancellations, gateway errors, response truncation, and token usage to help diagnose issues. Your API key is never logged.
Privacy & data
Your API key is stored locally using VS Code's SecretStorage. Chat prompts and context are sent directly from your editor to the LLM Gateway endpoint (https://api.llmgateway.io/v1) to generate responses. No data passes through any other third party.