Understand AI leaderboards

Learn what benchmark scores measure and what they leave out, look up any unfamiliar word, read a real leaderboard, then try running a model on your own computer.

5 steps · about 16 min of reading

1. Step 1

   What is an AI benchmark?

   A plain-language guide to AI benchmarks, what the scores mean, and why a high score doesn't always mean a model is right for you.

   3 min read

2. Step 2

   How to read AI benchmarks

   What AI benchmark scores actually mean (percent correct, arena ratings, pass@1), which ones matter for everyday use, and the caveats to keep in mind.

   4 min read

3. Step 3

   AI glossary

   Short, plain-language definitions of the AI words you'll see in news and benchmark results, from tokens and context windows to pass@1 and Elo.

   6 min read

4. Step 4

   Read the AI benchmarks page

5. Step 5

   Run AI on your own computer (free and private)

   How free apps like Ollama and LM Studio run AI models on your own computer, what your computer needs, and what to expect.

   3 min read