Highlights:

  • Strawberry will feature a pricing model limiting the number of messages users can send per hour, unlike the free or subscription-based GPT models.
  • Strawberry has an intentional delay that enables it to process each query more thoroughly, reducing the likelihood of errors.

According to a recent source, OpenAI’s advanced AI model, Strawberry, may launch ahead of schedule, with its release potentially happening within the next two weeks.

The model was initially set to launch in the fall. Still, according to the media house, two sources familiar with the situation have indicated that the release is now being expedited.

Last month, sources reported that OpenAI was developing the Strawberry model, highlighting its unique emphasis on reasoning. Unlike OpenAI’s current models, which respond almost instantly, Strawberry will take between 10 and 20 seconds to process and respond to questions or prompts.

This intentional delay enables Strawberry to approach each query more carefully, minimizing the likelihood of errors. By taking additional time to process responses, OpenAI aims for Strawberry to perform better, particularly in tasks requiring complex or multistep reasoning. According to anonymous sources cited by the report, the model is designed to excel in areas such as solving math problems, coding, and handling business tasks like generating comprehensive marketing plans.

The report states that Strawberry will be integrated into ChatGPT as a standalone option within the platform. While the exact method of access is still unclear, one possibility is that users may be able to choose Strawberry from a list of available models in the ChatGPT interface.

The report suggests that Strawberry might adopt a different pricing model. Strawberry is expected to impose limits on the number of messages users can send per hour, unlike current GPT models, which are available for free or via subscription. There could also be a premium tier for users seeking faster response times. Existing ChatGPT subscribers are likely to receive early access to Strawberry before it becomes available to free users.

Despite the claims of Strawberry’s enhanced reasoning capabilities, sources noted some limitations with the new model. At launch, it will only handle text-based queries, unlike GPT-4o, which can respond to user-uploaded images thanks to its multimodal features. Additionally, some early testers have reported that Strawberry occasionally takes too long to respond to simpler questions, and the perceived improvement in response quality may not always justify the extended wait time.

The report also noted that while Strawberry is designed to retain past conversations for more personalized responses, it sometimes has difficulty doing so.

As with any new AI model, some initial challenges are expected, but Strawberry’s potential became evident when it was revealed to be a key factor in OpenAI CEO Sam Altman’s brief ouster last year.

Shortly before Altman’s dismissal, Strawberry, then referred to as “Q,” was flagged as a significant concern by some OpenAI researchers due to its advanced capabilities. Reports at the time indicated that these researchers were worried Q marked a major leap toward achieving “artificial general intelligence” and that its development was being hastily pushed forward.

AGI, or artificial general intelligence, is a sophisticated type of AI that can understand, learn, and apply knowledge across a range of tasks, much like humans. There are concerns that AGI systems could develop in unpredictable ways, akin to Skynet from the “Terminator” films, leading to unforeseen and potentially harmful consequences.

Regardless, OpenAI appears determined to move forward with Strawberry, and its launch could provide the company with new momentum in the competitive race to develop increasingly powerful AI models.