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

improved .py tool, final_answer and changed model to gpt-4.1-mini

Initial commit with LFS-tracked files