Gemini OpenAI Proxy is a small open-source gateway that converts the OpenAI API protocol into Google’s Gemini protocol, so tools built for ChatGPT’s API can talk to Gemini models without code changes. It has about 722 stars on GitHub and was last updated in October 2025.
Many applications, chat front-ends and agent frameworks speak the OpenAI chat completions format. This proxy sits in front of Google Gemini and translates incoming OpenAI-style requests into Gemini API calls, then translates the responses back. That includes multimodal requests in the style of GPT-4 Vision, which are mapped to Gemini Pro Vision. You run the proxy yourself, point your existing OpenAI-compatible client at its endpoint, supply a Gemini API key, and your software effectively gains Gemini as a backend. It is a focused, single-purpose utility: protocol translation, done simply, under an MIT license.
In practice it is often paired with chat front-ends and tools that only expose an OpenAI endpoint setting. Because it is self-hosted, requests pass only between your server and Google, with no third-party intermediary in a position to log prompts or responses.
The proxy is completely free and open source. Your only cost is Gemini API usage billed by Google under your own key, plus whatever small server you run the proxy on.
It is for developers who have an app, chatbot front-end or agent stack built against the OpenAI API and want to try or switch to Gemini models without rewriting integrations. It is also handy for evaluating Gemini side by side with OpenAI models through one interface.