# Cog

> Cog is an open-source tool for building and running machine learning models in containers, making it easy to package and deploy ML models consistently.

Cog is an open-source tool that packages machine learning models into standard, production-ready containers. It eliminates the complexity of Docker and environment configuration by automatically generating container images from a simple configuration file. Cog ensures reproducibility and consistency across development and production environments, making ML model deployment straightforward for data scientists and engineers alike.

- **Containerized ML Models**: *Automatically builds Docker containers for your ML models without requiring Docker expertise.*
- **Simple Configuration**: *Define your model's environment in a single `cog.yaml` file specifying Python version, dependencies, and GPU requirements.*
- **Standard HTTP API**: *Cog automatically generates a standard HTTP prediction API for your model, ready for deployment.*
- **GPU Support**: *Seamlessly handles CUDA and GPU dependencies, ensuring your model runs correctly on GPU-enabled hardware.*
- **Reproducible Environments**: *Pins all dependencies and system packages so your model runs the same way everywhere.*
- **Open Source**: *Freely available under an open-source license on GitHub, with community contributions welcome.*

To get started, install Cog via the CLI, create a `cog.yaml` configuration file in your project directory, define your predictor class in Python, and run `cog build` to generate a container image. The resulting image can be run locally or deployed to any container-compatible cloud platform.

## Features
- Automatic Docker container generation for ML models
- Simple YAML-based configuration
- Auto-generated HTTP prediction API
- GPU and CUDA support
- Reproducible environments
- Python dependency management
- CLI for building and running models
- Open-source and community-driven

## Integrations
Docker, Python, CUDA, GPU hardware

## Platforms
CLI, API, LINUX, MACOS

## Pricing
Open Source

## Links
- Website: https://lab.puga.com.br/cog
- Documentation: https://lab.puga.com.br/cog
- Repository: https://github.com/marciopuga/cog
- EveryDev.ai: https://www.everydev.ai/tools/cog