Run GPT-5 through Codex Key in 5 minutes

Codex Key is an OpenAI-compatible gateway that gives you GPT-5.4, GPT-5.5 and Codex without an OpenAI account, VPN, or waitlist. You pay per token — card or crypto.

What you get

• 3 models on one key: gpt-5.4, gpt-5.5, codex
• OpenAI-compatible API — works with Cursor, Cline, Continue, Codex CLI, any SDK
• Real-economy packs: ~47M tokens for $2.50, ~1.5B for $45, ~3.4B for $90

Step 1. Sign up

Open codexkey.ru/en and sign up — email or Telegram. No email confirmation; the cabinet is available instantly.

Step 2. Top up balance

Three plans available:

Plan	Price	Tokens	$/1M
Starter	$2.50	~47M	$0.053
Pro	$45	~1.5B	$0.031
Team	$90	~3.4B	$0.027

Card payments clear instantly. Crypto (USDT TRC-20) confirms in 1–2 minutes.

Step 3. Get your API key

After topping up, open API Keys → Create. Copy the sk-clb-xxxxxxxxxxxxxxxxxxxxxxxx value. The key is shown once — save it to a password manager.

Step 4. Configure Cursor

In Cursor → Settings → Models:

• OpenAI API Key: paste your sk-clb-xxxxxxxxxxxxxxxxxxxxxxxx
• OpenAI Base URL: https://codexkey.ru/v1
• Enable gpt-5.4, gpt-5.5

Same applies to Cline, Continue, Codex CLI — anywhere you can set OPENAI_BASE_URL.

Step 5. First request via curl

curl https://codexkey.ru/v1/chat/completions \
  -H "Authorization: Bearer sk-clb-xxxxxxxxxxxxxxxxxxxxxxxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [{"role": "user", "content": "Hello, GPT-5!"}]
  }'

You'll get a standard OpenAI response with a usage block. Tokens debit per tariff coefficient.

FAQ

Do I need a VPN?

No. Codex Key works from any IP — we route requests through transport nodes with clean ASN ranges.

Can I use this in commercial products?

Yes. The Team plan is sized for production loads up to 3.4B tokens per month. For larger volumes — message support.

Which models are supported?

GPT-5.4 (×1 coefficient), GPT-5.5 (×4.5), Codex / GPT-5.4-mini (×0.9). Fast and Priority modes — ×2. Details in the instructions.

What about privacy?

Requests aren't logged with content. We keep only metadata (model, tokens, latency) for billing. Full details in our privacy policy.