The latest version of OpenAI's large-scale language model "GPT-4" is likely to appear next week. At the event "AI in Focus - Digital Kickoff" held in Germany on March 9, Andreas Braun, CTO, Lead Data & AI STU of Microsoft Germany, said that GPT-4 will be announced next week. It revealed that. German tech media "heise online" reported.


Brown calls GPT-4 a "game changer." “Next week we will introduce GPT-4. A multimodal model will appear, offering completely different possibilities, for example video.”

Multimodal input (prompts) using images and voices is gaining attention.

On March 3, Microsoft announced a new AI model "Kosmos-1" called Multimodal Large Language Model (MLLM). Kosmos-1 supports images as well as words, and has the ability to add explanations to images and answer questions related to images.

There are also reports that Microsoft plans to incorporate GPT-4 into the search engine Bing.

GPT-4 is rumored to have more than 100 trillion parameters, compared to GPT-3's 175 billion parameters.