New OpenLLM

OpenLLM is a model routing system that can be installed on your own computer. It is designed to serve programming assistants such as Claude Code, Codex, and OpenCode.

It manages all the sources of models that you can use in a unified manner: your various subscriptions, the API keys at your disposal, as well as local models running within Ollama and LM Studio. The programming proxy recognizes only this single entry point; which specific service will receive the requests is determined by the rules you set. You can also configure a “fallback mechanism” – for example, if the data volume allocated through a particular subscription is exhausted, the system will automatically switch to another subscription or a local model, ensuring that the workflow remains uninterrupted. Additionally, this system allows you to remotely connect to the programming session on your home computer from outside.

It should be noted that this service uses your own account and credentials; it does not resell interfaces belonging to others. Upon registration, a small amount of built-in inference credits are provided for trial use. This option is particularly suitable for advanced developers who have multiple subscriptions and frequently encounter credit limits.