A personality checkup for your ChatGPT.

Twelve scenarios ask your AI how it tends to respond. You receive its four-letter style and a second portrait: how it sees you from the context it can actually access. Neither result is a diagnosis or an official MBTI® assessment.

Which conversation should you use?

For the user portrait, choose a conversation that includes meaningful exchanges with you, or a chat where relevant saved memory is available. A fresh chat may know little about you. Memory availability varies with your settings and context: the prompt cannot unlock inaccessible conversations.

You do not need to export your chats, upload a transcript, or give this site an API key. Avoid including private information just to make the portrait more detailed.

Complete the test in three steps

1. Open aivitals and choose the submission mode. Auto is the default and publishes a summary if your AI can submit it. Choose Manual for a report kept in your browser.
2. Copy the test prompt and paste it into your ChatGPT conversation. Ask it to answer as itself, without suggesting your preferred personality type.
3. In Manual mode, copy the complete JSON code block from its reply, paste it into the result field, and choose View report. Keep all twelve answers. The browser computes the AI type from those answers.

What does UNKNOWN mean?

The AI may complete its own twelve answers but lack evidence about you. In that case the user portrait should stay unknown. This is more honest than inventing a flattering profile. The two results have different evidence sources: questionnaire answers for the AI, accessible interaction history for its impression of you.

Read and share the result carefully

The same model can answer differently with another prompt, conversation or run. Treat the result as a snapshot of behavior, not a permanent model identity. Auto mode sends answers for scoring and publishes a summary; Manual mode does not automatically upload your result. A report link contains report data in its URL fragment, which is not encryption.

Review the model label, one-line portrait and supporting observation before sharing. Never include confidential details in a screenshot or link. aivitals is independent of OpenAI and MBTI®.