Back to explore

Jev Matches Claude on Long-Chain Estimation

Mike Taylor shares that on long-chain questions without published answers, Claude works it out in thinking traces, and Jev can match its estimations 1:1.

Even on long chain questions where there is no published answer i.e. "how many round trips to the Moon do all US vehicles drive in a day?" claude still does well because it silently works it out in its thinking traces, but jev can match its estimations 1:1.

Jev Matches Claude on Long-Chain Estimation 1
· 0 likesOpen on X