[yanosh_kunsh]

So does that mean that LLM inference could go down significantly in price and/or context length would dramatically increase?