GPT-4 is bigger and better than ChatGPT, but OpenAI won’t say why

“That’s something that, you know, we can’t really comment on right now,” OpenAI chief scientist Ilya Sutskever said when I spoke to the GPT-4 team at a video call one hour after the announcement. “It’s quite competitive there.”

Access to GPT-4 will be available to users who sign up for the waitlist and to paid premium ChatGPT Plus subscribers in a limited, text-only capacity.

GPT-4 is a large multimodal language model, which means it can respond to both text and images. Give it a picture of the contents of your fridge and ask it what you could make, and GPT-4 will try to find recipes that use the pictured ingredients.

“The continued improvements in many areas are remarkable,” says Oren Etzioni of the Allen Institute for AI. “GPT-4 is now the standard by which all foundation designs will be evaluated.”

“A good multimodal model has been the holy grail of many large tech labs over the past couple of years,” says Thomas Wolf, co-founder of Hugging Face, the AI ​​startup behind the large open-source language model BLOOM. “But he remained elusive.”

In theory, combining text and images could allow multimodal models to better understand the world. “It might be able to tackle traditional weak points in language models, like spatial reasoning,” says Wolf.

It is not yet clear if this is true for GPT-4. OpenAI’s new model appears to be better at some basic reasoning than ChatGPT, solving simple puzzles such as summarizing blocks of text into words starting with the same letter. In my demo, I was shown GPT-4 summarizing the OpenAI website announcement using words beginning with g: “GPT-4, Breakthrough Generational Growth, Gets Better Ratings. Safeguards, advice and gains made. Gigantic, revolutionary and globally gifted. In another demo, GPT-4 took a document about taxes and answered questions about it, citing reasons for his answers.