In the new interface, users can select specific model versions via a dropdown menu, including “Latest” (currently 5.4), 5.2, 5.0, or o3.

Additional settings are available through the “Configure” option. Users can, for example, enable the former auto-switching feature, allowing ChatGPT to move from Instant to Thinking mode when handling more complex queries.

OpenAI has also simplified the response regeneration menu and discontinued the “Nerdy” personality style. At the same time, the company is rolling out GPT-5.4 mini and improving GPT-5.3 Instant, which according to the changelog now uses less sensational language.

The routing system — where ChatGPT decides which model handles a request — has long been a point of criticism. Many users found it opaque at launch, as it did not always select the most capable model, leading to concerns that OpenAI might be routing expensive queries to cheaper models to reduce costs.