AI API key checker
Check whether an OpenAI, DeepSeek, Claude, Gemini or Qwen key works, which models it can see, and how fast it answers.
Your key stays in your browser.It goes straight to the provider, never to our servers, and we neither log nor keep it. Use a dedicated key with a spending cap, and delete it in the provider's dashboard when you are done.
When an integration fails with a vague error, the first thing to rule out is the key: has it been revoked, has the balance run out, can it reach the model you configured? Pick the provider, paste the key and the checker does two things. It asks the provider for the list of models the key can see, a request providers do not bill as usage. Then, if the model box is filled in, it sends one tiny completion asking for the word OK and reports the reply and the round-trip time — and that call does spend tokens, charged to your account like any other. It is the same idea as the 'channel test' button in songquanpeng/one-api (30k+ stars), the open-source gateway many teams use to pool keys from several providers behind one endpoint.
How it works
- Step one requests the provider's model list (for Claude, through Anthropic's official SDK). A 401 or 403 there stops the check; a provider without a list endpoint just moves on to step two.
- Step two, only when a model is named, streams a completion with the system prompt 'Reply with the single word OK.' and the message 'ping'; the reply, the latency and the number of models appear as results, with every model ID listed below.
- Nothing in the check passes through this site: the key is held in the page's memory unless you choose to remember it for this tab, it is sent only to the provider's own API, and the panel is masked from session recordings.
- Test with a key created for the purpose and delete it in the provider's console afterwards — any key that has been pasted into a web page is worth rotating, however much you trust the page.
Where your data goes
With your own key, what you type and the key go from your browser straight to the provider you choose; neither passes through Hysen Labs. In hosted mode, your text goes through our server to our provider (DeepSeek) and is billed in credits; we keep the token counts and cost of each run for billing, never the text itself or the answer. How the provider handles the text is governed by its own privacy policy.
This tool handles keys and credentials, so nothing about a run is saved, not even to your own history.
About your API key
We do not collect, store or leak your key: it lives only in this page's memory (unless you tick “remember in this tab”) and is gone when you close it. Still, treat any key you have pasted into a web page with care — create a dedicated key with a spending cap for use here, and delete or rotate it in your provider's dashboard when you are done.
What it costs
This tool is free, with no sign-in and no points.
Common questions
- What do the error codes mean?
- 401: the key is wrong, revoked, or meant for a different provider or regional endpoint than the one selected. 403: the key is valid but is not allowed that model or endpoint — it is not enabled for your account, project or region. 429: you are rate-limited, or the quota or prepaid balance is used up; OpenAI reports an empty balance this way. Any other status is shown with the provider's own message — DeepSeek, for instance, answers 402 when the balance is insufficient, and a 404 usually means the model name is misspelled.
- What does a network error mean?
- The browser never got an HTTP answer. Your connection or a proxy blocked the request, the provider is unreachable from your network, or the provider refuses calls made from web pages (CORS) — in which case the key may be fine but cannot be tested in a browser. Try it from a terminal instead: curl https://api.openai.com/v1/models -H "Authorization: Bearer $KEY". A custom OpenAI-compatible endpoint also fails this way when its base URL is wrong or it sends no CORS headers.
- Does the check cost money?
- The model list is not billed as usage; the test completion is. On an ordinary chat model it is a few dozen tokens, a tiny fraction of a cent at typical prices. A reasoning model may think before replying 'OK' and bill a few hundred tokens. Clear the model box to run only the free half.
- Why does the key work here but not in my app?
- The check proves that the key, the model name and the provider work from this browser. If your app still fails, look at what differs: the base URL (a regional or proxy endpoint), the exact model ID, an organisation or project header, or your server's outbound network. Copying the model ID from the list shown here rules out the most common typo.
The open-source behind it
This tool is a self-contained implementation. songquanpeng/one-api (MIT) does the same job as a library — if you need this behaviour inside your own program, start there rather than calling a web page.
songquanpeng/one-apiAlso known as
- openai api key checker
- api key tester
- check if openai key is valid
- deepseek api key test
- claude api key check
- llm api key validator