Meta appears to be closing the gap with OpenAI. According to a report surfaced by MSN (via Bing News), Meta's forthcoming AI model — codenamed "Watermelon" — has reached the same performance level as OpenAI's GPT-5.5 on key benchmarks.
The claim reportedly comes from Alexandr Wang, who leads Meta's Superintelligence effort. Wang reportedly told employees that the coming model matches OpenAI's flagship, according to the report.
A separate item carried by AOL.com (via Google News) frames the same development plainly: Wang says Meta's coming AI has "caught up" with OpenAI's flagship model.
A few things are worth keeping in mind. The performance claim is described as reported, not independently confirmed, and it originates from Meta's own leadership rather than an outside evaluator. Benchmarks — the standardized tests used to compare AI systems — are one measure of capability, but strong benchmark scores do not always translate into a better everyday experience for users. The available sources also do not specify a release date for Watermelon, how the benchmarks were run, or which specific tests were used.
Why it matters: if Meta has genuinely pulled level with OpenAI's top model, it signals that no single company has a durable lead in the race to build the most capable AI — a contest that shapes which products, tools, and assistants millions of people will eventually use.