304 North Cardinal St.
Dorchester Center, MA 02124
OpenAI announced the new GPT-4 language model, which the company says is “more creative and collaborative than ever before.” She “can solve complex problems with greater accuracy due to greater knowledge and problem-solving abilities.”
OpenAI is already working with some companies to integrate GPT-4 into their products. In particular, the language model will be used by Duolingo, Stripe and Khan Academy. In addition, GPT-4 will be available to ChatGPT Plus subscribers via the API.
During normal communication, the difference between GPT-4 and GPT-3.5 may not be noticeable, but when solving more complex problems, the differences become apparent. GPT-4 is a multimodal language model, which allows it to process incoming information in both textual and graphical form. By parsing text and images simultaneously, the language model can interpret more complex input data. In the examples below, the model explains memes and unusual images: