OpenAI has reinstated access to its GPT-4o model in ChatGPT for paying subscribers, just a couple of days after replacing it with GPT-5. The reversal comes following complaints from users who felt the newer model delivered less satisfying responses.
The company billed its new GPT-5 model as its “smartest, fastest, most useful model yet”, featuring a new system that uses a “real-time router” that switches between a lightweight model for simple queries and a deeper reasoning model for more complex tasks. OpenAI claimed it would generate more engaging and relevant answers, but feedback from some ChatGPT users suggested otherwise.
On Reddit, many reported that GPT-5’s replies were shorter and less accurate compared to the previous model. One common complaint was that the newer model’s answers felt brief and dry, lacking the personality and detail users valued.

OpenAI CEO Sam Altman acknowledged these concerns, stating on X that GPT-5 would “seem smarter starting today” and promising improvements. He also pledged greater transparency about which model is responding to a query, along with increased usage limits for Plus subscribers. The newer model will remain as the only option for free users, however.