FYI, due to Llama's naming scheme, there is no such thing as Llama 3.2 405B. 8B/70B/405B models are either Llama 3, 3.1, or 3.3 (except for 405B which wasn't initially released), while Llama 3.2 only contains 1B, 3B, 11B (vision), and 90B (vision) models. It's a bit confusing.
Meta both improves the technology or inference, while also trapping themselves alongside every other person training models to always update the training set every few months, so it knows what its talking about with relevant current events
