I finally ran both models side by side and the difference was obvious

I spent last weekend testing GPT-4o and Claude 3.5 Sonnet on the same coding task - converting a messy CSV file into a clean database. GPT-4o got lost halfway through and started hallucinating column names that didn't exist. Claude handled it on the first try without any hand-holding. Has anyone else found one model way better for data cleanup work?

2 comments

2 Comments

jenkins.alice11d ago

Same here, Claude crushed it while GPT messed up the data structure. What kind of CSV were you working with?

thomasm4110d ago

Oh yeah, mine was a mess of nested quotes that even I couldn't untangle (and I wrote the darn thing).

-1