GPT-5 and Codex API for Cursor, Codex CLI, and VS Code
Get an OpenAI-compatible API key, top up a prepaid balance, and track every request by model, tokens, cost, status, and errors.
How It Works
From registration to a working Codex setup in a few minutes.
Get your key
After registration codexkey issues a personal API key.
Connect Codex
Prepared instructions for VS Code, Codex CLI, Cursor, Zed, JetBrains, OpenCode, OpenClaw.
Track usage
See balance, request count, token spend, logs and model usage in the cabinet.
Build cheaper
Buy larger token packs to reduce the effective cost per 1M tokens.
For developers
Minimal setup for any OpenAI-compatible client.
- Base URL
- https://codexkey.ru/v1
- Key
- Authorization: Bearer sk-clb-...
curl https://codexkey.ru/v1/chat/completions \
-H "Authorization: Bearer sk-clb-..." \
-H "Content-Type: application/json" \
-d '{"model":"gpt-5.4","messages":[{"role":"user","content":"Say hello"}]}'Codex CLI, VS Code, Cursor, Zed, JetBrains, OpenCode, OpenClaw
Prepaid balance: usage is charged by actual prompt/output/reasoning tokens; cost and errors are visible in the cabinet log.
For billing and diagnostics, Codex Key stores technical request metadata: time, key, model, status, tokens, cost, and failure error.
Focused on AI coding workflows
Codex Key does not try to replace every gateway or observability platform. It covers the practical path: personal OpenAI-compatible key, prepaid balance, token logs, and fast Cursor/Codex CLI/VS Code setup.
Direct provider accounts
Best when you already have accounts, cards, and infrastructure for every API provider.
OpenRouter / Vercel AI Gateway / Requesty
Strong global multi-provider gateways. Codex Key is simpler for quick AI coding onboarding: cabinet, key, balance, and per-request cost visibility.
Portkey / Helicone
Great for observability and production tooling. Codex Key focuses on access, prepaid balance, and readable token logs without extra infrastructure.
ProxyAPI / NeuroAPI / GPTunneL
A closer class for API access. Codex Key is tuned for AI coding workflows, OpenAI-compatible clients, and IDE setup.
Models and compatibility
Use the familiar OpenAI-compatible format: pass the model in the request, and review available options and multipliers in the cabinet.
- GPT/Codex families for coding and agent workflows
- Chat Completions API for OpenAI-compatible clients
- Multipliers, errors, and cost are visible in request logs
Token Packs
Top up your balance. Larger packs give a better effective price per 1M tokens.
- Everything in Start
- Priority support
- Better effective rate
- Everything in Base
- For constant development
- Maximum savings per 1M
Example: 10M tokens in coding tasks are charged from the prepaid balance by actual prompt/output/reasoning tokens. Exact cost depends on the model, mode, and cabinet multipliers.
We do not store your chats.
Your conversation history stays locally in your client. codexkey does not maintain a separate database of chats; request processing goes through the configured API flow.
FAQ
Common questions before starting.
How does it work?
You get a cabinet and a personal API key. Requests go through codexkey using the configured endpoint and your token balance.
Do I need my own server?
No. You only need to update your Codex/client configuration. Prepared scripts and manual docs are available.
Which IDEs are supported?
The setup is designed for VS Code, Codex CLI, Cursor, Zed, JetBrains, OpenCode, OpenClaw, and OpenAI-compatible clients.
How are tokens charged?
Charges are based on real request usage. Some models or priority modes may apply multipliers, visible in the cabinet.
What if my balance runs out?
New requests stop until you top up. Your cabinet, key, logs, and history remain available.
Can another user see my key or logs?
No. Keys, balance, request usage, and logs are tied to each user account.
Start using Codex AI through codexkey.
Open your cabinet, copy your API key, top up tokens, and connect your client in minutes.