LLM Leaderboard

1. Smartest — highest Artificial Analysis Intelligence Index.
2. Best for coding — highest Artificial Analysis Coding Index.
3. Best for agents — highest Artificial Analysis Agentic Index.
4. People's choice — highest Arena Elo on at least 5,000 votes, never an effort variant.
5. Frontier on a budget — cheapest blended price within 5 index points of the smartest.
6. Grunt work — cheapest blended price at three quarters of the smartest model's index.
7. Fastest that still thinks — most output tokens per second above that same three-quarters floor.
8. Newest at the frontier — most recently released of the top ten by index.
9. Best open weights — highest index among models whose weights are published.
10. Open-weights gap — how long ago the closed frontier reached the best open model's score.

Open weights means OpenRouter lists a Hugging Face id for the model, closed means it lists the model without one, and a model it does not list is unknown. Cards are picked from one row per model, batch tiers and announced rows left out, and the filter bar does not move them.