Soju06 / codex-lb

项目摘要

Codex-lb is a Python-based load balancer and proxy for pooling multiple ChatGPT/Codex accounts with usage tracking, a web dashboard, and OpenAI-compatible endpoints. It distributes API requests across accounts using configurable routing strategies while monitoring token usage and costs.

为什么值得关注

It solves a practical problem for power users and teams with multiple ChatGPT accounts who want to maximize usage limits, track costs, and maintain session locality. With 2,200+ stars, 77 contributors, and 64 releases in 6 months, it has strong community traction and active development.

优势

• Multiple routing strategies (capacity-weighted, round-robin, fill-first, etc.) for flexible traffic distribution
• Comprehensive usage tracking with per-account token/cost monitoring and 28-day trends
• OpenAI-compatible endpoints work seamlessly with Codex CLI, OpenCode, and any OpenAI client
• Active development with frequent releases, CI/CD, Docker support, and a large contributor base

局限性

• No examples directory provided, which could slow onboarding for new users
• Relies on pooling multiple ChatGPT accounts which may violate OpenAI's Terms of Service
• 46 open issues suggest some ongoing stability or feature gaps

使用场景

• Developers with multiple ChatGPT Plus/Pro accounts wanting to maximize API usage across all of them
• Teams pooling AI API access and tracking per-member usage and costs
• Power users wanting detailed visibility into their AI API consumption patterns
• Users of Codex CLI or OpenCode wanting automatic failover and load distribution across accounts