Artificial Analysis
artificialanalysis.ai
Independent benchmarks comparing AI model quality, inference speed, and pricing across providers — updated continuously with real measurements rather than self-reported figures.
Why it is useful
Cuts through marketing claims by publishing actual latency, throughput, and output quality measurements across all major models and providers. When choosing between a fast cheap model and a slower capable one for a specific task, this is where to start. The price-per-token comparisons alone save hours of spreadsheet work when evaluating API providers.