ProxyPool Hub: Use Claude Code, Codex CLI, Gemini CLI for Free with Multi-Account Pooling

#ai #programming #opensource #devtools

What is ProxyPool Hub?

ProxyPool Hub is an open-source, multi-protocol AI API proxy server that lets you pool multiple free accounts to use popular AI coding assistants — without paying for API keys.

It supports:

Claude Code (Anthropic format)
Codex CLI (OpenAI Responses API)
Gemini CLI (Google Gemini API)
OpenClaw (Multi-provider)

Why?

Free-tier AI accounts have usage limits. When one account hits its limit, you have to wait. ProxyPool Hub solves this by rotating across multiple accounts automatically — when one is rate-limited, it seamlessly switches to the next.

Key Features

Multi-account pooling — Add multiple ChatGPT / Claude accounts, auto-rotate on rate limits
API key management — Bring your own API keys (OpenAI, Anthropic, Gemini, etc.) as fallback
Smart routing — Account-first or API-key-first priority, sticky or round-robin rotation
Protocol translation — Anthropic to OpenAI format conversion, so any client works with any backend
Visual dashboard — Web UI for account management, usage analytics, request logs, and model mapping
One-click setup — Configure Claude Code, Codex CLI, Gemini CLI, or OpenClaw with a single button
Free model access — Route haiku/small models to free providers (no account needed)

Quick Start

npm install -g proxypool-hub
proxypool-hub

Then open http://localhost:8081 to access the dashboard.

Architecture

The proxy receives requests in any supported format, translates them to the appropriate upstream API, and streams responses back — all transparently.

Claude Code / Codex CLI / Gemini CLI / OpenClaw
              |
       ProxyPool Hub (localhost:8081)
              |
    +---------+---------+
    |         |         |
 ChatGPT   Claude    API Keys
 Accounts  Accounts  (fallback)

Dashboard

The web dashboard provides:

Real-time account status and token expiry tracking
Usage analytics (per account, per model, daily/monthly)
Request logs with full request/response inspection
Model mapping configuration
One-click CLI configuration for all supported tools

DEV Community