GPT-5 on the horizon? Sam Altman’s cryptic tweet with ‘Strawberry’ fuels speculation

OpenAI has been working on Project Strawberry for some time, and Altman’s latest tweet has sparked rumours around GPT-5.

Sam Altman took to his X account to post a cryptic tweet with an image of strawberries. (Express Image/Reuters)

In July, it was reported that OpenAI was working on its approach to AI models under a project code-named ‘Strawberry’. While OpenAI has been tight-lipped about the project, from whatever there is in the public domain, Strawberry models aim to enable AI models to not only answer but also plan and navigate the internet autonomously. The capability of ‘Project Strawberry’ is reportedly termed as ‘deep research’ by OpenAI. 

Now, weeks after Project Strawberry made its way to the news feed, OpenAI CEO Sam Altman on Wednesday shared a tweet that has been construed as a confirmation that OpenAI is indeed working on ‘Project Strawberry’. Altman’s tweet on X (formerly Twitter) read, “I love summer in the garden’. His post also has an image showing a pot of strawberries in a private garden. 

i love summer in the garden pic.twitter.com/Ter5Z5nFMc

— Sam Altman (@sama) August 7, 2024

Incidentally, Project Strawberry was earlier known as Project Q* (read Q-Star), an AI model with significant advancements in terms of reasoning capabilities. Following his tweet, several users reacted with different theories. 

A user, Robert Maciejko, an entrepreneurial leader, retweeted Altman’s post and said, “If the past is any guide, this testing is ahead of an important launch. Sam Altman posted this odd strawberry post. Strawberry is apparently the code name for Q*, which is rumoured to be the AI Ilya saw before he fired Sam.”

IS THE AI THAT GOT SAM ALTMAN FIRED? GPT 5?

There’s a new model on https://t.co/jTep96FhWs called “anonymous chatbot,” apparently from OpenAI. I tried it, and it outperformed Gemini 1.5 Pro, the current leaderboard champion, on a simple prompt of mine. You can try it there too… https://t.co/rZCjS3Ql2A

— Robert Maciejko (@4Maciejko) August 7, 2024

Maciejko also highlighted how advanced abilities like autonomous navigation of the internet have sparked debates. He said that autonomous internet navigation, advanced voice, and vision capabilities, Sora for image and video generation, and the new offering from OpenAI could be called ChatGPT-5. 

Meanwhile, another user Moses Senpai said that it was August and there was no news about GPT-5. He added that GPT-4 was getting dumber and OpenAI co-founders were joining rival companies or starting their own firms. “Rivals are getting better models out (3.5 Opus later this year and Llama 4) Meanwhile Sama is posting about gardening,” wrote Moses. 

It’s august .. there hasn’t been news about GPT-5 ..

Gpt -4 keeps getting dumber ..

openAI cofounders are joining rivals or starting their own ai firms.

Rivals are getting better models out (3.5 opus later this years and llama 4)

Meanwhile Sama is posting about gardening https://t.co/U7KUyfL56Y

— Moses (@MosesSenpai) August 7, 2024

Another AI enthusiast, Ethan Mollick commented saying that OpenAI was the only company whose corporate communication strategy consists of ‘obscure hints, Delphic pronouncements and riddles that could fit into adventure games. 

OpenAI is the only company whose corporate communication strategy consists of obscure hints, Delphic pronouncements, and riddles that could fit into adventure games.

(Strawberry is rumored to be an advanced reasoning approach in their new models: https://t.co/fwAChGhH9B) https://t.co/7p4LBcZ3ds

— Ethan Mollick (@emollick) August 7, 2024

Bilal, another X user, shared his views by suggesting that OpenAI has GPT-5 but it is undecided about the release till the US elections owing to fears of ‘cataclysmic change’. In his long quote tweet, Bilal said that the competition is catching up and GPT-4o is underwhelming leading to dissatisfaction among power users and developers. He also said that there was a likely increase in subscription cancellations. 

Here’s my fan fiction rant on what’s been going on.

OpenAI has GPT-5. They have determined they can’t release it because they think it’s going to cause some cataclysmic change. This is despite the overwhelming historical evidence of them always being wrong about this. I guess… https://t.co/AAhn1v7sxZ

— Bilal (@deepwhitman) August 7, 2024

Bilal thinks that OpenAI is trying to maximise GPT-4o’s potential through various tactics and marketing strategies. However, the internal conflicts and leadership changes suggest a struggle over strategy. According to him, the situation seems like a power struggle with increasing pressure from both internal and external forces.