Skip to main content
Back

codex-lb Lets Developers Pool ChatGPT Accounts to Beat Codex Rate Limits

codex-lb lets you pool multiple $20 ChatGPT accounts behind a single OpenAI-compatible endpoint, with a dashboard, usage tracking, and smart routing strategies

Repo, Full Access
  • codex-lb is an open-source load balancer that pools multiple ChatGPT accounts behind a single OpenAI-compatible proxy endpoint.
  • It solves Codex's aggressive rate limits by rotating across several $20/month Plus accounts automatically, with no manual switching.
  • Eight routing strategies (capacity-weighted, round-robin, fill-first, etc.) are switchable live from the dashboard with no restart needed.
  • Supports Codex CLI, OpenCode, OpenClaw, and the OpenAI Python SDK out of the box via compatible endpoints.
  • Deployable via Docker one-liner or uvx codex-lb; Kubernetes Helm chart and PostgreSQL backend also supported.
  • The repo has 2,400+ stars, 370+ forks, and 68 releases -- actively maintained by a large contributor community.

OpenAI's Codex has a rate limit problem. Usage has grown faster than expected, and a consistent pattern has emerged: developers find real value, then hit a wall. The message is blunt: "You've hit your usage limit. Upgrade to Pro, visit the Codex settings to purchase more credits, or try again at Apr 8th." Being blocked for hours or days breaks any productive workflow. The community's answer is codex-lb, a self-hosted load balancer that pools multiple ChatGPT accounts behind a single proxy endpoint so your tools keep running.

The project started from a practical frustration. One user on X explained it directly: "I have 5 Codex subs right now. The $20 ones. And I use codex-lb to balance between them. I never need to logout, connect to the next one, etc." That post was enough to push the repo past 2,400 stars, with over 370 forks and 60+ contributors now actively maintaining it.

What it does

codex-lb sits between your applications and your ChatGPT accounts, handling account management and usage controls without requiring direct OpenAI API access. It distributes requests across multiple ChatGPT accounts automatically, tracks tokens, costs, and 28-day trends per account with real-time updates, and works with Codex CLI, OpenCode, and any OpenAI-compatible client.

The proxy authenticates as your ChatGPT accounts, not as an API client. You need active ChatGPT subscriptions, not API keys. This keeps you inside the consumer subscription model, which is where Codex's agentic coding features actually live. Codex typically costs around $100–$200 per developer per month at normal usage, so pooling several $20/month Plus accounts is a meaningful cost reduction for teams that need sustained throughput.