Agentic work costs ~input^2. For a single message, they cost the same. For long conversations, you end up paying much less.

I would not be surprised if it’s the same method we use for finding important stuff that we can’t remember where we put: “All right where would I put it this time to make sure I find it in the future? Ah there it is!”

LLMs are future predictors after all.

We also don’t know if there were ather message boards we don’t know of. Or was it from the same time as the other board? They don’t need to find the same channel every time, just some of the time.

# Flash, off-peak

    cache-hit 2.5x
    cache-miss 1.57x
    out 2.36x

# Flash, peak

    cache-hit 5x
    cache-miss 3.14x
    out 4.71x

Edit: fixed the numbers and formatting