Comment by stingraycharles

Fairly certain that model sizes and computational requirements will grow as the price for LLM compute drops.