AI video trends are the patterns in what creators are publishing: which visual styles are spreading, which models they choose and which ideas get the most engagement. NeoMundo's trend radar at /trends measures these patterns from the prompt gallery, a collection of more than 4,400 public AI images and videos collected from creators on X, each with its prompt, model, category, ratio and engagement.
The main number is share growth. For every style, the radar calculates what share of new works it represents in the current window and compares that with the previous window of equal length. As an illustration, if a style made up 2% of works in the previous 14 days and 4% in the last 14 days, its share doubled. Share is used instead of raw counts because the total number of works changes from week to week, and share growth separates a style that is truly gaining ground from one that simply rose with the whole gallery.
Model share works the same way but is reported as a change in percentage points. If one model goes from 20% to 25% of new works, that is a change of +5 points. This shows how attention moves between models such as Seedance, Kling, Veo and Hailuo on the video side, or Nano Banana and GPT Image on the image side.
Treat these AI trends as signals, not guarantees. A short window reacts fast but can be noisy; a 30-day window is steadier. The most useful pattern is a style or model that rises across the 7, 14 and 30-day views at the same time, which suggests a durable shift rather than a single viral post.
