OpenAI Codex CLI — Accessible from China, Pay-Per-Token

OpenAI Codex is a capable coding agent, but accessing it from Mainland China requires a VPN, and billing requires ChatGPT Plus or foreign credit cards. Router One is a drop-in OpenAI-compatible endpoint with stable access, per-token billing, and smart routing for automatic failover.

Stable from China

api.router.one reachable from China Telecom, Unicom, Mobile — no VPN, no dropped connections mid-task.

Pay-per-token, no subscription

Skip ChatGPT Plus. Top up with WeChat Pay, Alipay, Stripe, or USDT — pay only for what Codex actually uses.

Smart routing failover

EWMA-based latency routing with automatic failover when upstream providers degrade. Your Codex sessions stay online.

Configure in 30 seconds

Point the OPENAI_BASE_URL environment variable at Router One before launching Codex CLI.

terminal

# Use Codex CLI through Router One
export OPENAI_BASE_URL=https://api.router.one/v1
export OPENAI_API_KEY=sk-your-router-one-key

Read the Codex CLI alternative guide

FAQ

Do I still need ChatGPT Plus?

No. Router One is pay-per-token with no subscription. You only pay for what Codex actually consumes per request.

Will my existing Codex configuration work?

Yes — just change OPENAI_BASE_URL to https://api.router.one/v1 and use your Router One API key. All existing Codex CLI flags and behaviors remain unchanged.

What about latency from China?

Router One's edge infrastructure is optimized for Chinese ISP routing, typically delivering 80–150 ms TTFB to api.router.one without a VPN.

Can I set budget caps?

Yes. Set per-project and per-API-key spending limits in the dashboard with real-time token counting and alerts.

Start using Codex CLI in China

Get started free