Meta has postponed the release of its highly anticipated Behemoth large language model (LLM) until the fall, according to a report from The Wall Street Journal. Originally slated for an April debut during Meta's inaugural AI conference, LlamaCon, the model’s release was first delayed to June before being pushed back again.
Behemoth is intended to be Meta’s most advanced LLM yet, positioned to serve as a “teacher” model for future generations of AI. However, sources within Meta told the Journal that engineers remain unconvinced the model provides a significant improvement over Llama 4, which was released in April. That perceived performance gap is reportedly behind the decision to delay.
The delay adds pressure on Meta, which has publicly committed to becoming a global leader in AI. The company has already integrated its Meta AI assistant across Facebook, Instagram, WhatsApp, and Messenger, and recently launched a standalone Meta AI app, including features tied to its Ray-Ban smart glasses.
Katelyn Chedraoui, CNET’s social media and AI reporter, noted that this development fuels concerns about Meta falling behind rivals such as OpenAI and Google. “The race to build advanced and affordable AI is extremely competitive,” she said. “Meta risks losing momentum while competitors continue to push ahead.”