GPT-4 is a large OpenAI model from 2023 that reads both text and images and answers in text.
In plain words
GPT-4 was OpenAI's step up from GPT-3.5. Like earlier GPTs it is a transformer pre-trained to predict the next token, followed by an alignment stage that made it more factual and better behaved. Unlike GPT-3, it could take pictures as well as text as input. On many exams it scored like a strong human test taker, though OpenAI stressed it was still less capable than people in many real situations. OpenAI retired the original gpt-4 API model on 23 October 2026.
A quick example
In OpenAI's tests on a simulated bar exam, GPT-4's score placed it near the best tenth of human candidates. GPT-3.5, tested the same way, landed near the lowest tenth.
Where you'll meet it
- ChatGPT and the OpenAI API, starting in 2023.
Related
Sources
- PaperGPT-4 Technical Report, OpenAI (arXiv) · read 29 Sept 2026
- DocsGPT-4, OpenAI · read 29 Sept 2026
- DocsDeprecations, OpenAI API docs · read 12 Nov 2026
Checked by WhatIsAI against the sources above · 12 Nov 2026. Spotted an error? Report it.