OpenAI to Release More Powerful Generative AI... "GPT-5 Expected as Early as This Summer"

OpenAI is reportedly planning to release its next large language model (LLM), GPT-5, as early as this summer, according to a report by the U.S. economic media Business Insider citing sources on the 20th (local time).

Sources familiar with the matter stated that GPT-5 is being demonstrated to some corporate clients with the goal of launching within a few months. They also explained that GPT-5 is evaluated as "materially much improved" compared to previous models. However, the exact release schedule is undecided and may change.

GPT-5 is an upgraded version of GPT-4, which was released in March last year. OpenAI has previously stated that GPT-4 demonstrated "human-level capabilities" in the U.S. mock bar exam, the college entrance qualification test SAT, and math exams.

OpenAI is also known to have demonstrated a GPT bot powered by GPT-5 to some companies. Recently, OpenAI collaborated with the robotics startup 'FigureAI' to unveil 'Figure01,' a humanoid robot that converses and performs tasks like a human.

The media also reported that OpenAI is preparing to launch a new feature called 'AI agents,' which can autonomously perform tasks together with GPT-5.

Sam Altman, CEO of OpenAI, met with Korean startups at the San Francisco headquarters on the 14th and said, "I don't know exactly when GPT-5 will be released or how good it will be, but I expect significant progress in the next model."

He added, "I can confidently say that the next model (GPT-5) will be a leap forward," and mentioned, "It would be a mistake to think that the level of advancement in GPT-5 will not be high." Recently, he also appeared on a podcast hosted by American computer scientist Lex Fridman and said, "We plan to release an amazing model this year."