ChatGPT Strawberry: OpenAI about to launch a new model?

While ChatGPT-5 has attracted the interest of many analysts ahead of its release, it is the model called Strawberry that is currently making headlines. Strawberry stands out from other generative intelligence models by trying to offer “reasoned” answers to user queries. Thus, for the first time, where most current models offer results in just a few moments, Strawberry will take its time to respond: it will take 10-20 seconds for the model to offer a thoughtful answer to the user's question.

Everything to avoid mistakes

This is one of the pitfalls of artificial intelligence today: nothing forces it to provide a correct answer. Worse, it will make mistakes regularly. Strawberry aims to tackle this pitfall directly and try to provide the most correct answers possible and avoid errors as much as possible. By giving the model additional thinking time, OpenAI hopes to allow it to consider all the solutions to choose the best possible one. For now, Strawberry is designed to mainly solve problems related to mathematics, coding or even professional questions. It is a slower but more expansive model, which will be integrated into ChatGPT as a standalone option according to The Information.

Despite these good intentions, TheInformation's report reveals that ChatGPT Strawberry is still affected by some limitations: first, the model is only accessible in text version. Second, some testers have noted that the model takes a while to answer very simple questions. Finally, while Strawberry should normally remember previous conversations with the user, early testers of the model have noted inconsistencies in this area.

Strawberry ChatGPT was originally scheduled to roll out this fall, but OpenAI's new model is expected to be available in the next two weeks, much sooner than expected. This could be a way for OpenAI to stay in the conversation while it waits to release ChatGPT-5.