OpenAI could be about to drop Project Strawberry in huge ChatGPT upgrade

Rumors are circulating that OpenAI is about to drop a new model this week and it could be our first sighting of the highly anticipated “ Project Strawberry ” — a reasoning-focused AI.

Trying to follow artificial intelligence topics on social media over the past few days brought back memories of strawberry picking with my parents as a child. Every account has been dropping pictures of the fruit — including OpenAI CEO Sam Altman .

Altman posted a photo of a strawberry plant with the caption "I love summer in the garden". The picture revealed four strawberries, hinting that there might be a new version of GPT-4 built specifically for reasoning . This could run alongside GPT-4o, built for creation and interaction .

The first details of Project Strawberry leaked in May. It is a breakthrough in the road to artificial general intelligence (AGI) where AI can think and act like a human. It isn’t clear whether this will be integrated into an updated GPT-4 or be an entirely new model.

What is Project Strawberry?

No information on Strawberry has been released by OpenAI officially but documents have leaked showing it is the same as Q*. This is a process of training a model where an agent can make decisions by estimating the value of an action.

According to a leak shared with Reuters in May Strawberry has advanced reasoning capabilities in a way previous models have struggled. For example, it would allow ChatGPT to solve complex problems independently and verify its output.

An AI model with reasoning capabilities could revolutionize scientific research, enhance decision-making in complex fields like healthcare and finance, and accelerate technological innovation.

Even at a user level, there will be improvements. It could offer more optimized travel plans, and health recommendations and even analyze your spending habits and create a fully customized budget to help you on the path to financial freedom.

OpenAI says there are five levels on the path to AGI and we're currently on level one — chatbots. This is AI where you can have a natural conversation but without much in the way of problem-solving capabilities. Strawberry is level 2 — reasoners.

Isn't this all just hype?

On X the user iruletheworldmo started randomly posting Strawberries and suggesting we're on the cusp of a 'level two' breakthrough.

The user wrote "welcome to level two. how do you feel? did I make you feel?" and Altman replied "amazing tbh". This triggered a flurry of Strawberry-related — then Altman tweeted a picture of the strawberry plant, further fueling the hype train.

When you combine this with the fact Google has its Made by Google event this week and OpenAI has a habit of making big announcements every time Google holds an event — and the hype train went into overdrive.

I don't know if we're about to see a new reasoning model. If we do see one I don't know what form it will take but it does feel like we're on the cusp of another leap in the power and capabilities of AI models.

More from Tom's Guide

• Meta is building a superintelligent AI — and one expert warns of ‘significant moral issues’
• Google DeepMind CEO claims we’re ‘just a few years away’ from human-like AI
• ChatGPT could rival human intelligence with next-level upgrade