OpenAI’s upcoming new artificial intelligence model Strawberry could be set to launch earlier than planned, with a prospective release slated for “within the next two weeks,” according to a report today by The Information.

The new model had originally been scheduled to launch during the fall, but The Information’s report cites two sources with knowledge of the matter as saying that its release is now being accelerated.

The Information first revealed that OpenAI was working on Strawberry last month, saying that it sets itself apart from other generative AI models with its focus on reasoning. As a result, it will generally take between 10 and 20 seconds to consider its response to any question or prompt, as opposed to the almost instantaneous responses of its existing models.

This is a deliberate delay that allows Strawberry to address each query in a more methodical way, thereby reducing the chance of errors. By taking time to think, OpenAI hopes Strawberry will deliver better responses, especially in tasks that require complex or multistep reasoning. The anonymous sources told The Information that the model is designed to excel in areas such as solving math problems, coding and performing business-focused tasks like creating a detailed marketing plan.

According to the report, Strawberry will be integrated within ChatGPT, but it will be a standalone option within that platform. It’s not clear exactly how users will be able to access it, but one possibility is that users will simply be able to select Strawberry from a list of available models within the ChatGPT interface.

Strawberry may have a different pricing model too, the report said. Unlike the existing GPT models, which are available for free and via subscriptions, Strawberry will have a pricing structure that imposes limits on the number of messages users can send each hour. There may also be a higher-priced tier for those wanting faster response times. Existing ChatGPT subscribers will likely gain early access to Strawberry before it becomes available to free users.

Despite the claims around Strawberry’s advanced reasoning, The Information’s sources said there are still a few limitations with the new model. For one thing, it will only be able to process text-based queries at launch, unlike GPT-4o, which has multimodal capabilities that allow it to respond to user-uploaded images. In addition, some early testers have reported that Strawberry sometimes takes too long to respond to simpler queries, and the perceived improvement in the quality of its responses may not justify the lengthy delays in getting them.

Another issue is that Strawberry, although designed to remember previous conversations so it can deliver more personalized responses, sometimes struggles to do this, the report said.

Still, as with any new AI model, there are always going to be a few teething problems, but the potential of Strawberry was made clear when it was revealed to be one of the main reasons behind OpenAI Chief Executive Sam Altman’s brief ouster last year.

Just prior to Altman’s sacking, Strawberry, then known simply as “Q,” was singled out as a major concern by certain OpenAI researchers, thanks to its advanced capabilities. According to reports at the time, the researchers were worried that Q represents a major breakthrough on the journey toward building “artificial general intelligence” and that its development was being rushed through.

AGI is an advanced form of AI that has the ability to understand, learn and apply its new knowledge across various tasks, similar to humans. Some fear that AGI systems could potentially take on a life of their own, similar to Skynet from the “Terminator” movies, resulting in some very unintended consequences.

In any case, OpenAI seems intent on going ahead with Strawberry, and its debut could give the company fresh momentum in the race among AI companies to create ever more powerful AI models.

Image: SiliconANGLE/Microsoft Designer