Guides to benchmarks, pricing and choosing a model, and notes on how ModelBench works.
ModelBench tracks 25 new releases, 12 price changes, and leaderboard moves for the week ending October 19, 2026.
Input vs output tokens, why output costs more, and how to estimate a monthly bill.
Time to first token, output speed and end-to-end time, measured ourselves through OpenRouter.
What ModelBench tracks, where each number comes from, and how to read the rankings without over-reading them.
A ranking is a list of point estimates. Intervals tell you which differences are real.