One Hub is an open-source API gateway and distribution system that presents many different LLM providers behind a single OpenAI-compatible interface. Forked from the popular one-api project, it adds support for more models, a richer statistics dashboard, and better function-calling support for non-OpenAI models. The project has about 3,000 stars on GitHub and is actively maintained.
Anyone reselling or centrally managing LLM access โ a team, a lab, or a small API business โ faces the same chores: issuing keys, tracking per-user quotas, load-balancing across provider accounts, and reconciling usage. One Hub packages all of that into a self-hosted web application. Administrators configure upstream “channels” for OpenAI, Claude, Gemini, Azure and many domestic Chinese providers; users then call one endpoint with one key, and the gateway routes, retries and meters every request. A built-in console shows consumption statistics, balances and logs, and features like token grouping and model mapping make it practical to run a tidy multi-provider service. Compared with the original one-api, One Hub focuses on broader model coverage and clearer analytics, which is why many operators migrating from one-api choose it as the successor for production deployments of shared LLM infrastructure.
One Hub is free and open source. You host it yourself and only pay for your own server and whatever upstream model APIs you connect to it.
One Hub is for teams, universities and independent operators who need to share, meter and manage access to multiple LLM providers through one gateway with proper accounting.