Can Claude make videos?
Yes, but not the way Sora, Veo or Kling do. Claude is a language model: it outputs text, including code. To make a video, it writes a program that draws each frame, usually with canvas, SVG, CSS or React, and a separate tool renders those frames into a video file with a headless browser and an encoder.
That is why Claude videos look like motion graphics: sharp type, shapes, charts and UI, rather than filmed scenes. It is also why they can be exact: the text and numbers are code, not guessed pixels.