Versions:
CC Desktop Switch is a lightweight desktop utility, published by CC Desktop Switch, designed to configure third-party Anthropic-compatible API providers for use with the official Claude Desktop client on Windows and macOS. Currently at version 1.0.26, with three recorded versions in its release history, the software allows users to manage providers such as DeepSeek, Kimi, Zhipu GLM, Alibaba Cloud Bailian, and Xiaomi MiMo, as well as custom third-party providers, and then apply the appropriate Claude Desktop third-party inference configuration with a single click. Unlike CLI-oriented alternatives such as farion1231/cc-switch, the tool targets regular desktop users by offering a simple graphical interface for provider setup, model mapping, health checks, and local gateway compatibility. Since version 1.0.18, Claude Desktop is configured to route requests through a local CC Desktop Switch gateway at 127.0.0.1, meaning the application must remain running in the background when third-party providers are in use; closing the window leaves it accessible via the system tray on Windows or through the hidden Dock app lifecycle on macOS. The local gateway keeps model mapping, protocol compatibility, extra headers, and upstream keys under local control. For safety and predictability, the tool exposes only explicitly mapped Claude-safe model routes in Claude Desktop and rejects unmapped Claude model routes rather than silently falling back to an internal default. Additional functionality includes importing Anthropic-compatible CC-Switch configurations, with OpenAI-format entries treated as opt-in, and performing provider connectivity health checks. Falling within the developer tools and AI assistant configuration category, CC Desktop Switch serves users who want to extend Claude Desktop beyond its default provider by connecting alternative Anthropic-compatible inference services without manually editing configuration files, while maintaining local control over credentials and routing behavior across both major desktop platforms.
Tags: