This leaderboard ranks AI video models by measured generation time: how long each model takes to return a finished clip for the same standardized prompts, timed across our evaluation runs rather than taken from provider marketing pages. Latency matters most for iteration-heavy workflows — look development, previsualization, and interactive products — where waiting minutes per generation breaks the creative loop. Cross-reference the overall leaderboard before committing: the fastest tier often trades away realism and prompt adherence, and the sweet spot is usually a fast variant of a top-scoring family.