OpenAI is set to release its next-level AI product named 'Strawberry' this fall, aiming to tackle advanced problems and tasks beyond the capabilities of current models like ChatGPT.

Strawberry is designed with advanced features that include solving previously unseen math problems, developing marketing strategies, and tackling complex word puzzles, such as the New York Times 'Connections' puzzle.

In internal tests, Strawberry has reportedly achieved over 90% on the MATH benchmark, significantly outperforming previous models like GPT-4 and GPT-4o.

To enhance its capabilities, OpenAI is utilizing the larger version of Strawberry to create synthetic training data, reducing reliance on traditional internet-sourced texts and images.

This synthetic data is expected to improve the quality of training and minimize AI errors, commonly referred to as hallucinations.

OpenAI is actively seeking new investors to support the ongoing development of its latest AI models, including Strawberry.

Concerns have been raised by OpenAI researchers regarding the risks associated with advanced AI systems like Strawberry, particularly in relation to artificial general intelligence (AGI) and its potential implications.

Ghazenfer Mansoor, CEO of Technology Rivers, emphasized the transformative potential of human-level reasoning AI across various sectors, especially in business.

The larger version of Strawberry is also being utilized to enhance the successor to GPT-4, which is codenamed Orion.

The anticipated launch of Strawberry is expected in the fall of 2023, although the exact date for the smaller, cost-effective version remains uncertain.

It remains to be seen whether this new version will enhance the performance of ChatGPT or GPT-4 this year, as OpenAI continues to innovate in the AI space.

Previously known as Q*, Strawberry was initiated by OpenAI's former chief scientist, Ilya Sutskever, and has undergone significant development since its inception.