← Glossary Term
Elo
A rating system borrowed from chess, now used to rank AI models by head-to-head comparisons.
Elo was invented to rank chess players. Everyone starts with a number, and it moves up or down depending on who you beat and how strong they were. Beating a much better opponent moves your rating a lot; beating a much weaker one barely moves it.
AI leaderboards use the same maths, comparing model outputs pair by pair, either judged by people or by another model. That makes Elo useful for questions with no single right answer, like which of two drafts is better written. The number only means something relative to the other models in the same pool, so an Elo score from one leaderboard cannot be compared with one from another.
Mentioned in
-
Nvidia's New Free Model Is Not the Smartest. It Is Just Very, Very Fast
-
xAI's New Image Model Lands Second in the Arena Rankings, and Brings Photoshop-Style Editing to Grok
-
Alibaba's newest model scores higher and guesses more: hallucination rate jumps from 23 to 40 percent
-
DeepSeek updated its cheap model and it now runs neck and neck with OpenAI's cheap model