> ModelPack: The Open Standard for Packaging AI/ML
Package AI/ML projects in industry standard OCI Artifacts. ModelPack brings vendor-neutral, container-like standardization to AI/ML, enabling automated pipelines to handle models, datasets, and code as unified, versioned packages.
The Problem
Packaging machine learning models is complex, often requiring teams to use proprietary package types or coble together open source tools. These inconsistent environments, manual processes, and proprietary formats lead to deployment failures, delays, increased operational costs, and vendor lock-in.
The Solution
ModelPack solves these challenges by providing a standardized, consistent, reproducible, and portable packaging format for AI/ML models, that is vendor neutral. This simplifies deployment, reduces errors, and ensures models work seamlessly across different environments.
How ModelPack Works
ModelPack defines a specification for packaging AI/ML models, including model files, dependencies, and metadata. This package can then be integrated into CI/CD pipelines for automated testing, validation, and deployment. The standardized format ensures that models are deployed consistently, regardless of the underlying infrastructure. ModelPack uses a simple YAML configuration to define your AI/ML package.
Built for CI/CD
Seamlessly integrate with existing CI/CD tools to automate testing, validation, and deployment, reducing manual effort and errors.
Works With Your Infra
Infrastructure-agnostic, ensuring consistent deployment on cloud, on-premise, or edge devices without compatibility issues.
Open Specification
The CNCF-governed open-source specification ensures consistency, enables community-driven improvements, and avoids vendor lock-in.
Learn more →Reference Implementations
The enterprise implementation of the ModelPack specification is KitOps, which provides:
- - A comprehensive CLI for packaging, pushing, and pulling ModelPacks
- - Python SDK (PyKitOps) for integration with notebooks and ML workflows
- - 30+ integrations with CI/CD and MLOps tools
- - Selective pulling of individual assets
- - Deployment capabilities
KitOps is a CNCF project, that has been adopted widely with over 150,000 downloads and is the most mature implementation of the ModelPack standard.
Learn more at kitops.org
Install KitOps
$ brew tap kitops-ml/kitops
$ brew install kitops
$ kit version
Why ModelPack Matters
Speed
ModelPack is essential for streamlining AI/ML deployment. By standardizing packaging, it reduces complexity, improves reliability, and accelerates time-to-market, allowing you to focus on building great models.
Flexibility
The AI/ML tooling market is changing daily - using open and established standards like ModelPack and OCI makes it easy for you to change tools without forcing a costly multi-month data migration.
Get Started Now
