Model Overview
Llama 4 Behemoth is Meta AI’s colossal 2-trillion parameter foundation teacher model, designed to synthesize distillation data and evaluate high-level emergent capabilities.
Postponement and Extended Training
While smaller Llama 4 family variants progressed through developer previews, public release of the full Behemoth weights was postponed to evaluate cluster stability, inference costs, and alignment bounds.
Current Status and Evidence
Formally acknowledged by Meta AI research communications, the model’s release precision remains unknown as engineering teams prioritize distillation efficiency.
Latest documented updates
Reports indicate public release of 2T parameter Behemoth teacher model is postponed.
Read source
Release timeline
Llama 4 Behemoth officially announced as teacher model in extended training.
Read source
Sources & verification
These sources support the historical milestone. Verification of an announcement does not imply current product availability.
- Meta AI introduces Llama 4 and previews 2T teacher model Official source
