RayBytes/ChatMock

What it solves

ChatMock provides a way to use ChatGPT models within third-party chat applications and coding tools that expect an OpenAI-compatible or Ollama-compatible API. It acts as a bridge, allowing users to sign in with their ChatGPT account and expose those models as a local server.

How it works

The tool runs as a local server (defaulting to http://127.0.0.1:8000) that mimics the OpenAI /v1/responses and Ollama endpoints. Once a user logs in with their ChatGPT account, ChatMock automatically discovers the available models and serves them to compatible apps via HTTP and WebSockets.

Who it’s for

Developers and power users who want to integrate their existing ChatGPT account and its specific models into their preferred IDEs, terminal agents (like Terax), or productivity tools (like Raycast).

Highlights

• API Compatibility: Supports both OpenAI-compatible and Ollama-compatible endpoints.
• Advanced Model Control: Configurable reasoning effort (from minimal to ultra) and thinking summary verbosity.
• Feature Parity: Supports tool/function calling, vision/image input, and web search capabilities.
• Performance Options: Includes a "fast mode" for supported models.
• Automatic Discovery: Automatically syncs and discovers models available to the signed-in account.