Replicate CLI & Cog

The Replicate CLI and Cog work together to streamline ML model deployment. The Replicate CLI runs and manages models on Replicate's serverless platform. Cog is a companion tool that packages models into production-ready Docker containers with automatic CUDA/cuDNN setup, OpenAPI schema generation, and local prediction testing. Together they provide a complete workflow from local development to cloud deployment.