Reddit community debates VLAs vs Nvidia world models

Reddit post cites Cosmos3 and Dreamzero papers. User asks about trajectory of vision-language-action models compared to world models and other approaches.