A week after GPT-5, the model picker came back
OpenAI's plan to let ChatGPT choose the right model for each question did not survive first contact with users.
Kivorent News desk, published , 1 min read
1 weekbetween GPT-5's launch and the picker's return
This is a look back at news from the week of . We wrote it on November 11, 2026, from the sources listed at the end.
When OpenAI launched GPT-5 on August 7, 2025, it removed the menu that let people choose a model. A router would decide for them, sending easy questions to a fast model and hard ones to a slower one.
Within a week the menu was back, TechCrunch reported, according to AI Weekly.
Why it mattered
The idea was sound. Most people should not need to know model names. In practice, users noticed when they were routed to a model that suited them less, and some had preferences the router could not see.
The episode showed how attached people become to the behavior of a particular model. It also shaped later launches. GPT-6 went to paying users first and to everyone else in a named, separate version.
Sources
- Issue #451: Generative AI trends 2025, AI Weekly, August 14, 2025
Spotted an error? Tell us and we will fix it.