How do I know if my OpenAI key is valid?

Paste it above. We call /v1/models with your key and report status, the list of models the key can access, and round-trip latency.

Does my key get logged anywhere?

No. The proxy forwards the key once, returns the response, and discards it. We do not log headers, request bodies, or the key value.

Can I tell if my key is a project key vs a user key?

Project keys start with sk-proj-. Service account keys start with sk-svcacct-. Plain sk-... keys are legacy user keys (still supported for now).

Why don't I see GPT-4o in my models list?

GPT-4o is gated by tier. New accounts can call gpt-4o-mini immediately; gpt-4o-full and o1 unlock once you've spent a small amount or topped up.

Is there a free tier?

OpenAI gives short-lived starter credits to new accounts. They expire after a few months. After that you pay-as-you-go.

How fast should an OpenAI key respond?

GET /v1/models typically returns under 300ms from a healthy region. Our latency benchmark runs 5 pings and reports p50 / p95 / p99 so you can spot regional slowdowns.

Why do I see a 401 even though the key is correct?

Whitespace. Paste-from-Slack and paste-from-1Password sometimes wrap the key in a zero-width character. Try retyping the last few chars or pasting into a plain text editor first.

openai

chat

embedding

image

audio

vision

Test your OpenAI API key.

Verify an OpenAI key in seconds, see which GPT-4o, o1, embedding, and DALL·E models you can call, and benchmark how fast OpenAI is responding to you right now.

Get a key API docs Pricing

Stateless proxy — keys never logged, stored, or persisted. What happens to your key →

API key·paste anywhere

Detected

OpenAI

What this key does

An OpenAI API key authenticates requests to OpenAI's REST API — chat completions, the Responses API, embeddings, image generation, audio transcription, and the moderation endpoint. Keys can be project-scoped or attached to a service account; both work the same way over the wire (Authorization: Bearer ...). Calling /v1/models is the cheapest way to confirm a key is alive without spending tokens.

How to get a OpenAI API key

Sign in at platform.openai.com and switch to the project you want this key tied to.
Open Settings → API keys and click Create new secret key.
Choose a permission scope (Restricted is safer than All) and a project.
Copy the key — you only see it once — and paste it here to confirm it works.
Set a usage cap on the project so a leaked key has a bounded blast radius.

Common errors and fixes

401 Unauthorized: Key is invalid, revoked, or pasted with extra whitespace. Generate a new key from the provider console and try again.
403 Forbidden: Key is valid but lacks permission for this resource. Check project / org / workspace scope, or that billing is set up for this key.
429 Too Many Requests: You hit the per-minute or per-day rate limit. Wait a moment and retry, or upgrade your tier.
404 Not Found: The endpoint or model id changed. Check the provider docs for the current path and model identifier.
5xx: The provider is having issues. Check their status page before assuming the bug is yours.

Security best practices

Store keys in an env var or secret manager — never commit them to a repo, even a private one.
Restrict scope: prefer per-project or per-deployment keys over a single root key shared across services.
Rotate on a schedule (90 days is a sane default) and immediately on suspected leak.
Audit usage in the provider console after rotation to confirm the old key has zero traffic.
Set per-key spend limits where the provider supports them, so a leaked key has a bounded blast radius.

Pricing at a glance

OpenAI bills per-million tokens for chat models and per-image for DALL·E. GPT-4o-mini is the cheapest reasoning-capable chat model in the catalog. Batch API requests are 50% off list price; cached input tokens are billed at a steep discount. See the live pricing page for current numbers.

FAQ

How do I know if my OpenAI key is valid?: Paste it above. We call /v1/models with your key and report status, the list of models the key can access, and round-trip latency.
Does my key get logged anywhere?: No. The proxy forwards the key once, returns the response, and discards it. We do not log headers, request bodies, or the key value.
Can I tell if my key is a project key vs a user key?: Project keys start with sk-proj-. Service account keys start with sk-svcacct-. Plain sk-... keys are legacy user keys (still supported for now).
Why don't I see GPT-4o in my models list?: GPT-4o is gated by tier. New accounts can call gpt-4o-mini immediately; gpt-4o-full and o1 unlock once you've spent a small amount or topped up.
Is there a free tier?: OpenAI gives short-lived starter credits to new accounts. They expire after a few months. After that you pay-as-you-go.
How fast should an OpenAI key respond?: GET /v1/models typically returns under 300ms from a healthy region. Our latency benchmark runs 5 pings and reports p50 / p95 / p99 so you can spot regional slowdowns.
Why do I see a 401 even though the key is correct?: Whitespace. Paste-from-Slack and paste-from-1Password sometimes wrap the key in a zero-width character. Try retyping the last few chars or pasting into a plain text editor first.

What this key does

How to get a OpenAI API key

Common errors and fixes

Security best practices

Pricing at a glance

FAQ

Test other providers

Related reading