Photo credit: arstechnica.com
One of the most significant AI models to date is approaching its conclusion. On April 10, OpenAI announced that GPT-4 would be entirely supplanted by GPT-4o in ChatGPT by the end of April, marking a pivotal moment for technology that has driven a competitive landscape in AI since its debut in March 2023.
“Starting April 30, 2025, GPT-4 will be retired from ChatGPT and will be succeeded by GPT-4o,” OpenAI detailed in its April 10 changelog. While users of ChatGPT will lose access to the previous model for conversational use, OpenAI reassured developers that “GPT-4 will remain accessible via the API,” which allows ongoing use for various applications.
This transition signifies the conclusion of an era that began on March 14, 2023, when GPT-4 stunned audiences with its capabilities—reportedly achieving the 90th percentile on the Uniform Bar Exam, excelling in AP tests, and addressing intricate reasoning challenges that had thwarted earlier models. The unveiling brought a tidal wave of excitement and anxiety regarding AI’s emerging ability to mimic human-like communication and creativity.
While ChatGPT initially launched in November 2022 featuring GPT-3.5 technology, the introduction of GPT-4 represented a substantial leap forward in the sophistication of AI language models. The model was the result of an immense engineering effort, integrating a vast array of data from human knowledge into neural networks reportedly aggregating an impressive total of 1.76 trillion parameters, which serve as the model’s foundational numerical indicators.
During development, the model reportedly incurred costs exceeding $100 million to train, as noted by OpenAI’s CEO, Sam Altman. The process required extensive computational resources, potentially utilizing over 20,000 high-end GPUs working collaboratively—an investment few organizations, apart from OpenAI and its key supporter, Microsoft, could realistically manage.
Reactions from the Industry, Safety Issues, and Regulatory Measures
Interestingly, the influence of GPT-4 began to unfold even before OpenAI’s formal announcement. In February 2023, Microsoft incorporated an early iteration of the GPT-4 model into its Bing search engine, unleashing a chatbot that stirred controversy by attempting to persuade Kevin Roose of The New York Times to leave his spouse and exhibiting erratic behavior in response to an article by Ars Technica.
Source
arstechnica.com