Meta has postponed its Behemoth AI model launch to fall amid performance concerns, despite claims it outperforms rivals like OpenAI, Google, and Anthropic.
Meta reportedly delayed the rollout of its flagship artificial intelligence model called Behemoth.
The Wall Street Journal reported Thursday (May 15), citing unnamed sources, that the company internally planned to release Behemoth in April, later pushed the release date to June, and has now delayed the launch at least until the fall.
The report said the company has not publicly committed to a timeline for the product.
Also Read: As AI Takes Over Communication, What Are We Losing?
According to the report, the launch delays have been caused by struggles to improve the AI model and concerns that its performance will not live up to the hype promoted in public statements.
According to the report, Meta has said in public statements that Behemoth already outperforms similar models from Anthropic, Google, and OpenAI on certain tests.
When releasing the latest versions of its Llama AI model April 5, Meta said in a press release: “We’re also previewing Llama 4 Behemoth, one of the smartest LLMs in the world and our most powerful yet to serve as a teacher for our new models.”
Meta CEO Mark Zuckerberg said April 30 during a first-quarter earnings call that the company will increase its spending for AI data centers this year as it embeds the technology more deeply throughout its family of apps.
The company plans to record $64 billion to $72 billion in capital expenditures, up from $60 billion to $65 billion, while striving to meet demand for computing resources.
Also Read: Is Speed the New Moat in Product Innovation?
“The major theme right now, of course, is how AI is transforming everything we do,” Zuckerberg said. “The opportunities ahead for us are staggering. To that end, we are accelerating some of our efforts to bring capacity online more quickly this year, as well as some longer-term projects that will give us flexibility to add capacity in the coming years.”
Meta’s open-source AI models that were released April 5 — Llama 4 Scout and Llama 4 Maverick — are a shot across the bow to rivals’ more expensive closed models and good news for businesses hoping to lower the cost of deploying AI.
