6 Questions Answered About OpenAI’s Model Router System Reversal

OpenAI has reversed course on a major feature for its free and lower-tier users, prioritizing speed over advanced reasoning capabilities as competition in the AI space intensifies.

Key Changes to ChatGPT’s Model Selection

OpenAI has quietly rolled back its automated model router for users on Free and $5-a-month Go tiers, according to a recent company blog post. Instead of the system automatically directing queries to more advanced reasoning models when needed, these users will now default to GPT-5.2 Instant, the fastest version of OpenAI’s new model series. While users can still access reasoning models, they must now select them manually.

The model router, introduced just four months ago alongside GPT-5, was designed to unify user experience by intelligently directing complex questions to more sophisticated AI models. CEO Sam Altman had previously noted that the router increased usage of reasoning models among free users from less than 1% to 7%, a change that likely increased OpenAI’s operational costs significantly.

Why the Reversal?

According to sources familiar with the matter, the router negatively affected OpenAI’s daily active users metric. While reasoning models can provide better answers, they take longer to respond—sometimes minutes for complex questions—and most consumers prefer speed over quality.

An OpenAI spokesperson stated that user feedback indicated Free and Go users preferred the default chat experience with the option to manually select reasoning capabilities when needed. The company also noted that its Instant models have improved and can now take more time to answer questions when necessary, narrowing the performance gap for most users.