Both help you choose AI video models, but they're different tools. WMArena is a focused, live, blind human-preference arena for world-model video generation. Artificial Analysis is a broad benchmarking suite spanning thousands of models across text, image, speech and video, combining automated metrics with a human arena. WMArena is depth on world-model video by human vote; Artificial Analysis is breadth across the whole model landscape.
Artificial Analysis is one of the most comprehensive AI benchmarking platforms — per-model pages, head-to-head comparisons, provider analysis, and named indices, across many modalities. Its video section includes an arena with blind side-by-side voting, an enforced minimum watch time, and Elo with confidence intervals. It's the place to see a model's standing across the entire field and on automated measures.
WMArena does one thing deeply: rank world models — starting with video — by blind human preference. Every battle starts from the same image and action, two anonymous models render the next-world clip, you vote, and a Bradley-Terry leaderboard updates. The question it answers is specific: given an action, which model renders what happens next most convincingly.
| WMArena | Artificial Analysis | |
|---|---|---|
| Scope | Focused: world-model video | Broad: text, image, speech, video |
| How it ranks | Live blind human preference only | Automated metrics + a human arena |
| Framing | World models — "given an action, what happens next?" | General model benchmarking |
| Best for | Human-judged world-model / video rankings | Breadth, automated metrics, cross-modality context |
| Method | Bradley-Terry from blind pairwise votes | Composite metrics + Elo arena |
Not really — they answer different questions. Artificial Analysis is excellent at breadth and automated measurement. WMArena is purpose-built for the thing automated metrics capture worst: whether a generated world looks right to a human and does what the action asked. For world-model video specifically, that human-preference signal is the one WMArena specializes in. See also WMArena vs LMArena.