Artificial Analysis

Artificial Analysis is an independent AI benchmarking platform that runs its own evaluations of language, image, video, speech, and embedding models rather than relying on labs' self-reported numbers. It's built for developers, AI teams, and enterprises deciding which model and inference provider best fit their use case, comparing options across intelligence, output speed, and price. Its signature Intelligence Index (currently v4.3) aggregates ten real-world evaluations into a single, regularly recalibrated score, which is what most differentiates it from single-benchmark leaderboards.