OpenAI's generative artificial intelligence (AI) 'ChatGPT' has achieved the highest ever daily active users (DAU) in Korea despite performance controversies and user complaints regarding the latest model 'GPT-5'. While the number of users temporarily decreased right after the release of GPT-5, OpenAI successfully rebounded with a quick response. In the global market, AI services from competitors such as Google, Meta, and xAI are rapidly catching up, but domestically, ChatGPT still maintains the widest user base.
According to the app statistics analysis platform Mobile Index on the 22nd, as of the 18th, the domestic app DAU for ChatGPT was recorded at 3,336,563. This is the highest figure since its domestic launch.
Earlier, ChatGPT showed a noticeable decline right after the disclosure of GPT-5 on the 8th. The DAU, which recorded 3.16 million on the 8th, dropped to 2.92 million on the 9th and further fell to 2.89 million on the 10th, reaching the 2.8 million range. This was interpreted as a result of user dissatisfaction with the cold and indifferent tone of the new model and the shorter answers, leading some users to leave. Among domestic and international users, reactions included comments like 'the friendly ChatGPT has disappeared' and 'the conversation has become stiff.'
In the existing ChatGPT app, users could select the model they wanted. Free users typically used GPT-3.5, while paid subscribers could choose advanced models such as GPT-4o or GPT-4.1. It was also possible to switch to engines with different conversation tendencies or response styles as needed. However, immediately after the release of GPT-5, the option essentially disappeared. Upon accessing the app, users were automatically assigned GPT-5 as the default model.
OpenAI seemed to have forced users to experience the new model. During this process, some users expressed frustration, saying they could no longer choose the familiar GPT-4o. In response, OpenAI quickly revised its policy to reopen the 'legacy model' menu, allowing users to select GPT-4o directly. GPT-5 also introduced a warmer personality mode and a safety feature suggesting breaks during lengthy conversations. Sam Altman, CEO of OpenAI, acknowledged that 'there is no single model suitable for everyone' and admitted that there were mistakes in the release process.
Subsequently, the number of domestic users rapidly recovered. After bouncing back to 3.24 million on the 11th, it maintained around 3.27 million on the 12th and 3.18 million on the 14th. In just a week, the count increased by about 500,000, reaching a record high of 3,336,563 on the 18th. This means that despite the initial exodus, it set a new record in a short period.
A similar trend is observed globally. Immediately after its release, GPT-5 stirred controversy over its completeness due to spelling errors and incorrect basic knowledge in the U.S. and Europe, but recent weekly users have surpassed 700 million, and paid subscribers have also exceeded 5 million. Particularly in emerging markets such as India, a low-cost subscription plan priced at $4.6 (approximately 6,400 won) has been introduced to rapidly expand its user base.
An AI industry official noted that 'after ChatGPT gained public recognition through the Ghibli profile photo craze, it quickly penetrated various areas including work, study, and daily life.' They added, 'except for specialized areas such as research or advanced development tasks, ordinary users have already formed the habit of using ChatGPT for everyday tasks like questioning, searching, and writing, so they are unlikely to easily leave.'
According to a report released by the Bank of Korea on the 18th, more than half of domestic workers (51.8%) are utilizing AI in their jobs, leading to an average reduction of 1.5 hours in weekly working hours. The Bank of Korea analyzed that the spread of generative AI may have raised Korea's GDP growth rate by up to 1 percentage point (P) since the end of 2022. Notably, Korea's AI utilization rate (51.8%) is double that of the United States (26.5%).