Meta’s next big AI brain, code-named Behemoth, just hit a major speed bump.

According to The Wall Street Journal, Meta has quietly pushed back the launch of its flagship AI model – again. Originally set to debut in April during Meta’s first AI dev conference, Behemoth was delayed to June… and now it’s looking more like a fall (or later) release. The reason? It’s just not ready.

Engineers are reportedly still fine-tuning the model and struggling to make it meaningfully smarter than its predecessors. That’s a tough sell in today’s turbocharged AI race, where Google, OpenAI, and Anthropic are dropping model upgrades like mixtapes.

Meta first teased Behemoth under its Llama 4 branding, calling it “one of the smartest LLMs in the world” and positioning it as a future AI teaching tool. But instead of unleashing the beast, Meta rolled out Llama 4 Scout and Llama 4 Maverick – lighter versions meant to hold the line while the big one cooks.

So for now, Behemoth remains in the lab. Meta seems to be taking the “don’t ship it ‘til it’s good” approach – which, honestly, is a pretty smart move in a market where first impressions stick and flops go viral.