Rumor on the ground is that they expected a much stronger model than this one.

Funny contrast with Anthropic. Ant does a "hero run," gets a model much more powerful than they expect. Meta does a hero run, gets a model much more mediocre than they expect. Read into this what you will, I guess?

Can you elaborate?

That's it. It's just a rumor. A model, which I don't even know of it's this one specifically, fell short of expectations. This rumor came up around mid March.

llama4 behemoth problems?