OpenAI was a pioneer in image/video generation, but later shifted its focus entirely to code generation to chase SOTA. Gotta admit, the Silicon Valley OGs like OpenAI know their stuff: images and videos grab attention and mindshare, while code is a productivity multiplier that makes them stronger and stronger Once the flywheel is spinning, they can plug image/video back into the workflow as modules. Of course, ByteDance is smart too. In China, the demand for image/video from DS hasn’t been fully met, so Douyin and Meitu are pulling in new users