Rumors are circulating that OpenAI will unveil a new model this week, which may be the first look at the long-awaited “Project Strawberry” - an AI focused on inference.

Following the topic of artificial intelligence on social media over the past few days has brought back memories of going strawberry picking with my parents as a child, and every account, including OpenAI CEO Sam Altman's, has posted pictures of strawberries.

Altman posted a photo of a strawberry plant with the caption, “I love summer in the garden.” The photo shows four strawberries, suggesting that there may be a new version of GPT-4 created specifically for inference. This could work in parallel with GPT-4o, which was created for creation and interaction.

The first details of Project Strawberry were leaked in May. This is a breakthrough on the road to artificial intelligence (AGI), where AI can think and act like humans. It is not clear if this will be integrated into an updated GPT-4 or if it will be an entirely new model.

No information on Strawberry has been officially released by Open AI, but documents have been leaked indicating that it is the same as Q*. This is the process of training a model that allows agents to make decisions by estimating the value of actions.

According to leaks shared with Reuters in May, Strawberry has advanced reasoning capabilities in ways that previous models have struggled with. For example, it allows ChatGPT to solve complex problems on its own and validate its output.

AI models with reasoning capabilities have the potential to revolutionize scientific research, enhance decision making in complex areas such as healthcare and finance, and accelerate innovation.

Improvements will also be seen at the user level. More optimized travel plans, health recommendations, and even analyzing your spending habits and creating a completely customized budget could help you on the path to financial freedom.

According to OpenAI, there are five levels on the path to AGI, and we are currently at level one, chatbots. This is an AI that can have natural conversations, but not much in the way of problem-solving ability. Strawberry is Level 2 - a reasoner.

On X, user iruletheworldmo randomly posted Strawberry, beginning to suggest that we are on the cusp of a “Level 2” breakthrough.

The user wrote, “Welcome to Level 2,” to which Altman replied, “Awesome.” Altman then tweeted a picture of a strawberry plant, adding to the hype.

This, combined with Google's Made by Google event this week and OpenAI's habit of making big announcements whenever Google holds an event, spurred the hype train.

We don't know if new inference models will emerge. But it feels like we are on the cusp of a new leap forward in the power and capability of AI models.