ChatGPT free users shifted to GPT-5.2 Instant as OpenAI scraps automatic model switching

ChatGPT users on the free tier and the lower-cost Go subscription will now interact with a simpler AI model by default after OpenAI revised its model selection system, ending automatic switching to more advanced models for complex queries, as per a report by Gizmodo. According to an update dated December 11 in ChatGPT’s release notes, OpenAI removed automatic model switching for reasoning tasks for free users and ChatGPT Go subscribers, a feature that earlier allowed the system to quietly route certain prompts to more powerful models when it determined deeper reasoning was required. Following the change, all prompts from free and Go users will be answered by GPT-5.2 Instant by default.

OpenAI informed that the update is intended to give users greater control over which model they use. While GPT-5.2 Instant will handle most everyday conversations, users who believe their query requires deeper reasoning can still manually select the more advanced Thinking model from the tools menu each time they want to use it. ChatGPT Go, which costs $5 per month and is available only in select regions, will continue to offer access to the Thinking model alongside free users, but it will no longer be triggered automatically in the background.

The company described GPT-5.2 Instant as a dependable option for routine tasks such as learning, writing and general queries, while positioning the Thinking model as better suited for more complex problems that require detailed and refined responses. The move follows earlier user frustration with OpenAI’s automatic model switching, particularly after the introduction of GPT-5, when frequent users said ChatGPT felt less expressive and more constrained. OpenAI chief executive Sam Altman later acknowledged the criticism, stating that the company was also dissatisfied with how confusing the model selection experience had become.

However, the change has also prompted questions around safety and performance. Previously, OpenAI said it automatically routed sensitive or emotionally complex conversations to its reasoning model because it handled such situations more effectively. With automatic switching now removed, that safeguard is no longer in place. OpenAI has stated that GPT-5.2 Instant has improved sufficiently to manage these conversations.