OpenAI has announced plans to end the use of the GPT-4 model in its ChatGPT chatbot. Starting April 30, the model will be completely replaced by GPT-4o, which is already the current default model. At the same time, GPT-4 will remain available to users via the OpenAI API.
In comparative evaluations, GPT-4o consistently outperforms its predecessor in text generation, programming, and STEM subjects. The latest updates to the GPT-4o model have improved its ability to follow instructions, solve problems, and conduct conversations, making it the natural successor to GPT-4.
The GPT-4 model was released in March 2023 for ChatGPT and Microsoft’s Copilot chatbot. It became OpenAI’s first model with multimodal capabilities, able to understand both images and text. According to OpenAI CEO Sam Altman, training GPT-4 cost over one hundred million dollars.
The discontinuation of GPT-4 is expected to be accompanied by the release of new models in ChatGPT. Among them are the GPT-4.1 family, as well as new task-solving models such as o3 and o4-mini.
