tiny decision models — one forward pass, no text generated
Text-to-video with synchronized audio
Automate tasks with n8n workflows