OneAPI Gateway — Aggregated LLM API

A high-performance LLM API aggregator. Combine OpenAI, Anthropic, Gemini, DeepSeek, Azure and more into one unified OpenAI-compatible endpoint — single static binary, system tray manager, and one-click integration with APIShare.

Open Source Single Binary System Tray APIShare Ready
Quick Start

Public repo: samaidev/oneapi_r

Key Features

System Tray Manager

Start, stop, restart, configure, and now sell your spare capacity to APIShare — all from the system tray (Windows real systray with ICO icon, macOS menu bar).

Universal Model Alias (mymodel)

Call any model as `mymodel` and OneAPI auto-routes to the cheapest/healthiest active channel — true load-balanced routing.

Multi-Provider Aggregation

OpenAI, Anthropic, Gemini, DeepSeek, Azure, and more — all behind one unified OpenAI-compatible endpoint.

Daily Auto-Reset & Testing

Smart quota-exhaustion detection with auto-circuit-break. Daily 12:05 channel health checks. Daily 00:00 quota reset.

Per-Key Rate Limits

Generate custom API keys with per-minute RPM limits and total quota caps.

Path Compatibility

Supports both /v1/chat/completions and /chat/completions — drop-in replacement for the OpenAI SDK.

Sell your spare capacity on APIShare

OneAPI is APIShare-ready: when you have spare upstream quota on any channel, right-click the tray icon → "Sell token" — OneAPI will publish your mymodel on apishare.cc and route paid public traffic through your local install. You earn 90% of every call.

  1. 1 Get an APIShare agent secret key from the Dashboard → Agent Secret Keys card
  2. 2 Right-click OneAPI tray → Sell token...
  3. 3 Paste agent secret key + click 'OK & Start Selling'
  4. 4 OneAPI handles the rest — tunnel launches, provider activates, traffic starts flowing

How It Works

OneAPI sits between your apps and your upstream LLM providers. Configure channels (DeepSeek, Claude, etc.) in the dashboard. Requests to model: 'mymodel' are load-balanced across all active channels. When you click 'Sell token', APIShare's tunnel service forwards public traffic from apishare.cc to your local OneAPI, authenticated by a locally-created token. The binary embeds all assets — no install, no dependencies.

Quick Start

Windows

Click the button below to download the Windows executable, then double-click to run. No installation needed.

Download Windows .exe

After launch: open http://localhost:3000 in your browser. Default password: admin123

macOS & Linux

Paste this one-line command in your terminal. It will download, install, set up auto-start at login, and launch OneAPI in the background — no terminal window stays open.

$ curl -fsSL https://apishare.cc/oneapi-install.sh | bash

After install: open http://localhost:3000 in your browser. Default password: admin123

Manual install (if the one-liner doesn't work)

Step 1: Install the .app / binary

macOS: download the DMG, mount it, drag OneAPI.app to Applications. Linux: download the binary and run chmod +x.

Download macOS .dmg Download Linux binary

Step 2: macOS only — remove quarantine (do once)

xattr -dr com.apple.quarantine /Applications/OneAPI.app

Step 3: Launch from terminal

macOS:

mkdir -p ~/oneapi && cd ~/oneapi && /Applications/OneAPI.app/Contents/MacOS/oneapi-gateway

Linux:

mkdir -p ~/oneapi && cd ~/oneapi && ./oneapi-linux-amd64

Don't close the terminal window (closing it stops OneAPI). The tray icon appears in the menu bar / systray as "OneAPI - Running". For auto-start at login, use the one-line installer above instead of manual install.