What is One API used for?

One API is an open-source LLM API gateway that unifies many model providers behind the OpenAI API format, with key management and redistribution. It is a good fit for managing API keys for several LLM providers and giving a team one OpenAI-compatible endpoint.

Is One API open source?

Yes. One API is open source under the MIT licence. Its source code is on GitHub at songquanpeng/one-api and is written mainly in JavaScript.

Yes. One API is open source, so the software itself is free to use.

Can I self-host One API?

Yes. One API can be self-hosted on your own server or infrastructure; there is no official hosted version.

What is One API an alternative to?

One API is an open-source alternative to OpenRouter and Portkey. Other open-source alternatives to OpenRouter include LiteLLM, New API and LLM Gateway.

Is One API actively maintained?

The most recent commit to One API was on 9 January 2026, and the latest release is v0.6.10, published on 2 February 2025. The project has 37k stars on GitHub.

One API: open-source alternative to OpenRouter, Portkey

About One API

One API is an LLM API management and key redistribution system that lets you reach many large language models through the standard OpenAI API format. It is distributed as a single executable and a Docker image, and the repository description is written in both Chinese and English with an English interface available.

It supports a long list of providers, including OpenAI and Azure OpenAI, Anthropic Claude, Google Gemini, Mistral, DeepSeek, Groq, Ollama, Cohere, xAI and many Chinese model services. Administrators can load balance across several channels, use stream mode, deploy on multiple machines, manage tokens with expiry, quotas, IP ranges and allowed models, generate redemption codes, group users and channels with different rates, retry failed requests and customize branding. A management API allows extension without code changes.

One API is licensed under MIT. The README warns that users must follow OpenAI's terms of use and applicable laws, and must not offer unregistered generative AI services to the public in China. It also instructs administrators to change the default root password after the first login.

Key features

OpenAI-format API for many model providers
Load balancing across channels
Token management with quotas and expiry
Redemption codes and user groups
Multi-machine deployment
Automatic retry on failed requests
Single binary and Docker image

Good fit for

→Managing API keys for several LLM providers
→Giving a team one OpenAI-compatible endpoint

Built with: JavaScript
Tags: llm; api-gateway; openai-api; key-management; proxy; self-hosted; docker; ai

One API: questions and answers

What is One API used for?: One API is an open-source LLM API gateway that unifies many model providers behind the OpenAI API format, with key management and redistribution. It is a good fit for managing API keys for several LLM providers and giving a team one OpenAI-compatible endpoint.
Is One API open source?: Yes. One API is open source under the MIT licence. Its source code is on GitHub at songquanpeng/one-api and is written mainly in JavaScript.
Is One API free?: Yes. One API is open source, so the software itself is free to use.
Can I self-host One API?: Yes. One API can be self-hosted on your own server or infrastructure; there is no official hosted version.
What is One API an alternative to?: One API is an open-source alternative to OpenRouter and Portkey. Other open-source alternatives to OpenRouter include LiteLLM, New API and LLM Gateway.
Is One API actively maintained?: The most recent commit to One API was on 9 January 2026, and the latest release is v0.6.10, published on 2 February 2025. The project has 37k stars on GitHub.