← sources

Are AI leaderboards even worth taking seriously anymore?

source: reddit (reddit:artificial)published 2026-11-15 14:07ingested 2026-11-16 02:34

retained text

Every time a new AI leaderboard drops theres suddenly a different model being called the best in the world and then you check another benchmark and the rankings are completely different. Gemini has been doing well in general chat rankings while Claude has been strong in coding, so apparently the best AI depends on which website you happen to open that morning.

Kalshi even has a market on which AI model finishes November at number one and it got me thinking about how much these rankings actually reflect what people use every day. I mean a model can score high on some benchmark and still give you a garbage answer to a basic question. At this point I’d rather hear from people who use these tools for actual work instead of another company announcing it beat everyone by three points on some test.



Has anyone here  switched their main AI because of these rankings or are we just watching companies fight over numbers that barely change the experience?

document fetched 2026-11-16 00:19 · canonical https://reddit.com/r/artificial/comments/1xwwdhv/are_ai_leaderboards_even_worth_taking_seriously