logoalt Hacker News

numbersyesterday at 6:31 PM2 repliesview on HN

I've stopped trusting these "trust me bro" benchmarks and just started going to LM Arena and looking for the actual benchmark comparisons.

https://arena.ai/leaderboard/code


Replies

stri8tedyesterday at 6:44 PM

I doubt this is representative of real world usage. There is a difference between a few turns on a web chatbot, vs many-turn cli usage on a real project.

nba456_yesterday at 6:40 PM

This is not any better of a benchmark