OpenAI is reportedly planning to release its next generation of artificial intelligence products in the fall.
The product, called “Strawberry,” will be able to solve problems and tasks that are beyond the capabilities of current AI models, Seeking Alpha reported on Tuesday (August 27), citing a paid article from The Information.
According to the report, Strawberry will be able to solve previously unencountered math problems, perform advanced tasks such as developing market strategies and solving complex word puzzles, as well as carry out “in-depth research.”
On July 16, it was reported that OpenAI was developing Strawberry, which aims to enhance the advanced inference capabilities of AI models, although exact details and timeline are unclear.
“OpenAI's pursuit of human-level reasoning isn't just a technological marvel. It's a story of pushing boundaries and creating new possibilities in every sector,” Ghazenfer Mansour, founder and CEO of Technology Rivers, told PYMNTS at the time. “For business, AI can dramatically change the way we manage supply chains, predict market trends with great accuracy, and make customer experiences hyper-personal at scale.”
As part of its latest product releases, OpenAI began rolling out advanced voice capabilities for its ChatGPT platform in July. The capabilities use the company's GPT-4o model to provide ultra-realistic voice responses, allowing users to have real-time, lag-free conversations with ChatGPT or interrupt mid-conversation.
The company also announced its new SearchGPT tool on July 25, offering prototypes to a limited test group. OpenAI said SearchGPT promises more concise and relevant results than traditional search engines, providing summarized answers with source links rather than pages of links.
In July, OpenAI also announced a smaller, more affordable version of its AI model ChatGPT. Dubbed GPT-4o mini and announced on July 19, this version replaces the previous model, GPT-3.5 Turbo.
“GPT-4o mini outperforms GPT-3.5 Turbo and other mini models on academic benchmarks for both text intelligence and multimodal inference, and supports the same range of languages as GPT-4o,” the company said in announcing the launch. “It also delivers strong performance in function calls, enabling developers to build applications that retrieve data or take actions in external systems, and offers improved long-context performance compared to GPT-3.5 Turbo.”
To stay up to date on all things PYMNTS AI, subscribe to our daily AI newsletter.
