OpenAI API key: platform.openai.com account, billing, secret key, and usage limits

Platform account → Billing → secret key → one smoke request → Limits and 429 basics.

You need an OpenAI API key for code and integrations—not the chatgpt.com web app alone. Official path: platform.openai.com account → Settings → Billing for metered usage → Settings → API keys to create a secret key → read Usage and Limits for RPM/TPM caps. Steps follow Quickstart, Authentication, and Rate limits.

The searchable job is OpenAI API key: platform.openai.com account, billing, create secret key, usage limits basics. Trust Billing, Usage, and Limits in your org over third-party quota spreadsheets. ChatGPT Plus does not include API quota.

Platform vs ChatGPT

Account, billing, and secret key

1. 01
   Open a platform.openai.com account

   Register with email or supported OAuth, verify email, and land in the Dashboard. Quickstart lists Sign up / Log in as step one. Use one org for your team and invite members under Settings if needed.

2. 02
   Set up Billing

   Go to Settings → Billing and add a supported payment method for your organization. API usage is metered by tokens (Pricing). Invoices and spend controls live here—not on chatgpt.com Settings.

3. 03
   Create a secret API key

   Settings → API keys → Create new secret key. Copy it into a password manager immediately. Authentication docs treat keys like passwords—never commit to Git or ship in frontend bundles. Use separate keys per app and revoke individually.

4. 04
   Smoke-test one request

   Export $OPENAI_API_KEY locally, then POST to https://api.openai.com/v1/chat/completions or use the Playground. Success means choices[0].message.content returns text and Usage increments.

5. 05
   Read usage limits basics

   The Rate limits guide explains RPM and TPM. Watch x-ratelimit-remaining-* headers. On 429, honor retry-after or lower concurrency. Caps also appear under Settings → Limits.

export OPENAI_API_KEY="sk-..."  # never commit

curl https://api.openai.com/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Say hello in one word."}]
  }'

Where setup usually breaks

401 Unauthorized

Missing, mistyped, or revoked key. Confirm Authorization: Bearer and Active status under Settings → API keys.

429 Too Many Requests

Hit RPM/TPM caps. Read headers and retry-after; Rate limits docs recommend exponential backoff.

403 or model not found

Your usage tier may not include that model. Pick a default from Quickstart or check Limits.

Billing not ready

Errors or zero usage despite calls—open Settings → Billing and confirm the org payment method and billing status.

Rotate a leaked key before debating models. One minimal request plus Usage confirms billing and limits faster than wiring a full IDE stack blind.