OpenAI Revolutionizes Model Selection in ChatGPT

OpenAI has recently revamped the way users select models in ChatGPT, adding a new dimension to the user experience. Instead of choosing from individual model names, users now see up to three distinct tiers based on their subscription: "Instant" for quick, everyday responses, "Thinking" for more complex tasks, and "Pro" for access to the most powerful models.

This new interface also allows users to select a specific version of the model from a dropdown menu. Available options include "Latest" (currently version 5.4), as well as earlier versions 5.2, 5.0, and o3.

For those seeking more advanced customization, additional settings are accessible under the "Configure" tab. This is where users can enable the old Auto feature, which allows ChatGPT to automatically switch from Instant to Thinking when the posed question is more complex.

OpenAI has also streamlined the repeat menu under the Responses tab and introduced a new personality style called "Nerdy." Additionally, the company has launched the GPT-5.4 mini model and made improvements to GPT-5.3 Instant, which now uses less sensationalist language, according to the changelog.

The routing system, which determines which model handles a given request, has long been a source of frustration for OpenAI. Many users found this system opaque, as the router did not consistently select the most performant model. This raised suspicions that OpenAI might be subtly directing costly requests to cheaper models to reduce computational costs.