switched model back to gpt-4o-mini due to poor performance / adding ytb videos due to block

Initial commit with LFS-tracked files

initial commit