Meta's next artificial intelligence model has reached performance levels matching OpenAI's GPT-5.5, according to internal remarks by Alexandr Wang, the company's superintelligence chief, as per a recent report by Business Insider.

Wang told staff at an internal town hall that the upcoming model, codenamed Watermelon, had closed the gap based on widely followed benchmarks, two sources familiar with the matter told the publication. The specific benchmarks were not disclosed.

Watermelon is still in training and uses "an order of magnitude more compute" than its predecessor, codenamed Avocado, which Meta released publicly as Muse Spark in April, Wang is reported to have said. If accurate, the assessment would mark the clearest indication yet that Meta's huge investments in chips, data centres, and top AI talent are beginning to deliver competitive results.

Meta has struggled to position its models at the industry's leading edge despite forecasting up to $145 billion in infrastructure spending this year. CEO Mark Zuckerberg appointed Wang last year to accelerate the effort, and the company has offered elite researchers hundreds of millions of dollars each to join.

OpenAI released GPT-5.5 in April and followed it with a more powerful GPT-5.6 late last month, though the newer model has not been widely released at the request of the US government. Meta declined to comment on the report, and OpenAI did not respond to a request for comment.