Free local diagnostic

Test your OpenAI API key and check model access

This free OpenAI API key checker verifies that your key works, fetches every model available to its project, and diagnoses missing models or restricted project permissions.

Use it when an app shows fewer OpenAI API models than expected or you need to confirm which models your key can access. The test runs locally in your browser, and BoltAI never receives or stores your API key.

Direct browser-to-OpenAI request
No key storage
No BoltAI proxy

Direct OpenAI check

Run a private model-access diagnostic

1. Your browser

Key lives in this tab

2. Direct request

api.openai.com only

3. Local report

Rendered in memory

This tool never writes the key to cookies, localStorage, sessionStorage, analytics events, or logs. It calls OpenAI directly without a BoltAI API route. Clear the diagnostic when you finish.

Where to check next

Review Model usage in OpenAI

This diagnostic shows what the key can currently access. To change that access, open the project that created the key and review its Model usage settings in the OpenAI dashboard.

Use a key you control

OpenAI recommends protecting API keys from browser exposure. Use this page on a trusted device. For a production key, consider creating a temporary diagnostic key and revoking it afterward.

  1. 01

    Select the affected project

    Open OpenAI Projects and choose the project where the API key was created. Make sure this is the same project used by the key you tested above.
  2. 02

    Open Project Settings

    In the OpenAI dashboard, open Project Settings for the selected project. Organization-level settings are a different scope, so confirm the project name before continuing.
  3. 03

    Choose Limits

    Select Limits in Project Settings, then scroll down to the Model usage section.
  4. 04

    Review Model usage

    Check whether the models you expect BoltAI to show are enabled for this project. Enable the missing models there, then run this diagnostic again. Permission changes may take a few minutes to appear.
  5. 05

    Refresh BoltAI’s model list

    In BoltAI, open AI Providers, select OpenAI, then choose Refresh Model List. If the website diagnostic shows more chat models while BoltAI still shows only a few, copy the safe report and send it to BoltAI support.

Prefer Terminal for a production key?

Keep the key in an environment variable and ask OpenAI for the same model list. The command prints model IDs only.

curl -s https://api.openai.com/v1/models \
  -H "Authorization: Bearer $OPENAI_API_KEY" | \
  jq -r '.data[].id' | sort

What the safe report contains

The copied report includes the HTTP result, request ID, model counts, diagnosis, and public model IDs. It excludes the key, organization and project IDs, owner values, and fine-tuned model IDs.

That gives support enough evidence to distinguish an OpenAI access restriction from a BoltAI filtering problem.

Frequently asked questions

Before you paste a key

Is it safe to enter my OpenAI API key here?
The test runs locally in your browser. Your browser sends the key directly to OpenAI, and BoltAI does not receive, upload, store, log, or share it. The key is not added to cookies, analytics, or browser storage.
How do I test whether my OpenAI API key works?
Paste the key into the checker and select Test API key. The tool asks OpenAI for the models available to that key. It then shows whether authentication succeeded, the returned models, and any error OpenAI reported.
How can I see which OpenAI models my API key can access?
Run the test above. The results contain every model returned by OpenAI and a separate list of models BoltAI recognizes as usable for chat. The result reflects the key and project at the time of the test.
Can this tell whether I have a personal or project key?
The key prefix can provide a format hint, but OpenAI does not reveal the key owner, account email, or definitive credential type. Project-prefixed keys are labeled as a hint, and the report explains this limitation.
Why are some OpenAI models not showing?
OpenAI may limit model access for the project that owns the key. BoltAI also removes known non-chat model families from its chat picker. Check Project Settings, then Limits, then Model usage in OpenAI and confirm you tested a key from the intended project.
Why does my OpenAI API key return only a few models?
A project model allowlist, an older project configuration, or a key from the wrong project can produce a very small model list. Review Model usage under the project Limits page, save any changes, and test the key again.
How do I change an OpenAI API key’s permissions?
Open the API key in the OpenAI project that created it and review its permission level. Key permissions and project model limits are separate controls, so also check the Model usage section under Project Settings and Limits.
Where do I enable models for an OpenAI project?
Open the affected project in the OpenAI dashboard, select Project Settings, choose Limits, and scroll to Model usage. Enable the required models there, save the changes, and refresh the model list in BoltAI.
Does the diagnostic make an AI request or spend model tokens?
The diagnostic only checks which models are available. It does not send a prompt, generate a response, or spend model tokens.
Why can’t the tool show my exact Model usage settings?
OpenAI does not include those settings when it returns the available models. Open Project Settings, choose Limits, and scroll to Model usage to review or change them.

Use your OpenAI models in BoltAI

Download the native Mac app and try it free. No account required.

Download BoltAI