Deno proxy dresses Gemini up as OpenAI for edge deploys

What it does

This project is a lightweight proxy that sits between your AI client and Google’s Gemini API. It translates requests into the OpenAI chat-completions format and pipes them through Deno, so standard tools like Cursor, Cherry Studio, or ChatBox can talk to Gemini without knowing the difference. The author targets users behind restrictive networks by offering a deploy-anywhere edge runtime that bypasses regional blocks.

The interesting bit

The whole thing is essentially protocol cosplay: Gemini speaks Google’s dialect, but the proxy makes it sound exactly like OpenAI. That lets you piggyback on Gemini’s generous free tier—including the 1-million-token context window—using off-the-shelf clients that expect an OpenAI endpoint. It is glue code, but useful glue for anyone who cannot reach Gemini directly.

Key highlights

• OpenAI-compatible API surface; drop-in replacement for ChatGPT clients
• Runs on Deno Deploy or self-hosted; no build step required
• Explicitly aimed at bypassing network blocks for users who need a direct line to Gemini from restricted regions
• Leverages Gemini 2.0’s free tier, including the 1M-token context model

Caveats

• The README itself warns that Deno’s services can be blocked, so the author actually recommends self-hosting on a foreign server rather than relying solely on Deno Deploy
• You still need your own Gemini API key from Google AI Studio; the proxy is just a pipe, not a key provider

Verdict

Worth a look if you’re in a region that can’t reach Gemini directly and want to plug it into standard OpenAI-shaped tools. Skip it if you already have unfettered Gemini access and don’t need another protocol translation layer.