Jozu Launches Enterprise Support for CNCF-Backed ModelPack and KitOps Standards
The ML ecosystem has long struggled with a fundamental problem: every tool has its own packaging format. Moving models between environments meant reformatting everything. Security scanning and governance were afterthoughts. Supply chain controls didn't exist.
Today, that changes. Jozu announces two developments that address these gaps: ModelPack and KitOps are now CNCF-backed industry standards for AI/ML packaging, and Jozu is launching comprehensive enterprise support for both projects.
Open Standards for ML Infrastructure
ModelPack establishes the vendor-neutral specification for how AI/ML artifacts should be packaged, versioned, and distributed in cloud-native environments. Created by Jozu in collaboration with Red Hat, PayPal, ANT Group, and ByteDance, ModelPack extends the OCI 1.1 standard to handle the large files common in ML work - models, datasets, and applications.
KitOps is the reference implementation of the ModelPack specification. It packages models, datasets, code, and configuration into cryptographically verifiable ModelKits. No proprietary formats. No vendor lock-in. Just reproducible deployments across any Kubernetes environment.
The numbers tell the story: 150,000+ downloads, production deployments at Fortune 500 companies and government agencies, and community-driven integration guides appearing across every major cloud platform.
Why Standards Matter
Traditional ML workflows suffer from disconnected stages. You might package your model one way for local testing, store it differently in a registry, and deploy it with yet another approach for production. Each transition introduces potential failures, security risks, and inconsistencies.
ModelPack and KitOps eliminate these risks by establishing a single, portable artifact that works everywhere. The same ModelKit you create during development is tested in staging and deployed to production - no translation, no reformatting, no surprises.
Because ModelKits use OCI standards, they work with existing container registries and Kubernetes infrastructure. No new clusters. No forced cloud migrations. Just standard tooling that integrates with your existing security controls, RBAC, and audit trails.
Enterprise Support from the Source
While the open source projects provide powerful capabilities, enterprises need guaranteed SLAs, expert guidance, and production-hardened features. Jozu now offers three tiers of enterprise support for both ModelPack and KitOps:
Professional Support includes business hours support with 48-hour response SLA, security patches and updates, and assistance with KitOps enterprise features.
Enterprise Support adds 24/7 priority support with 4-hour response SLA, dedicated Slack channel, quarterly health checks, custom CI/CD integration assistance, and hot fixes for emergency patches.
Enterprise Plus provides a dedicated support engineer with 1-hour response SLA, direct phone support, monthly health checks, quarterly architecture reviews, custom feature development, and influence on the ModelPack specification roadmap.
All support plans include assistance with both the base ModelPack standard and KitOps's enterprise enhancements - features like SHA verification, remote manifest inspection, cache management, registry authentication, and production-ready integrations with Jenkins, GitLab CI, GitHub Actions, OpenShift Pipelines, and 25+ other enterprise tools.
Beyond the Standard
KitOps provides critical enterprise capabilities that go beyond the base ModelPack specification. While any vendor can implement ModelPack compliance, KitOps is the only implementation currently deployed in enterprise production environments.
The project has been production-proven by global enterprises and security-conscious government agencies for over 18 months. This real-world usage has driven continuous improvements: optimized performance for large-scale deployments, robust CI/CD automation, comprehensive security scanning integration, and enterprise-grade reliability.
As Mohamed Nanabhay, Managing Partner at Mozilla Ventures, noted: "The AI/ML ecosystem has lacked open standards for packaging, versioning and sharing projects." ModelPack and KitOps fill that gap.
Real-World Implementation
Organizations are already using ModelKits to solve critical infrastructure challenges. One enterprise noted: "We're building a vendor-agnostic MLOps platform and KitOps ModelKits align perfectly with that vision. They work wherever our containers do - on-prem or in the cloud - giving us the freedom to store and deploy ML artifacts without being tied to specific infrastructure."
Another practitioner highlighted the governance benefits: "I really don't like the opinionated approach to how practitioners should log artifacts. I like the idea of using existing best practices in the MLOps space."
The technical pattern is straightforward:
- Package models, datasets, code, and configuration as OCI-compliant ModelKits
- Store in any container registry (Alibaba ACR, AWS ECR, Azure ACR, GCP GAR, Harbor, Artifactory)
- Deploy to Kubernetes using standard init container patterns
- Monitor and rollback using native Kubernetes tooling
Because ModelKits are just OCI artifacts, they integrate seamlessly with existing DevSecOps pipelines. Sign them with your container signer. Scan them with your security tools. Store them behind your firewall. Deploy them to air-gapped environments.
Looking Forward
Both ModelPack and KitOps are CNCF projects, ensuring vendor neutrality and long-term sustainability. The ModelPack specification is currently in the voting process to become the de facto CNCF standard for AI/ML packaging.
For organizations moving ML workloads to production, the challenge has never been just about models. It's about packaging everything needed to reproduce, test, and deploy those models consistently across environments. It's about security scanning, supply chain controls, and governance frameworks that basic registries don't provide.
ModelPack provides the standard. KitOps provides the production-proven implementation. Jozu provides enterprise support from the team that created both.
The tools are open source and available now. Enterprise support ensures you can deploy them at scale with confidence.
About Jozu
Jozu provides security-conscious organizations with AI/ML lifecycle tooling that ensures AI projects move efficiently from development to production without compromising compliance, security, or operational agility. As the creator and primary maintainer of both the ModelPack specification and KitOps, Jozu offers unmatched expertise in enterprise ML infrastructure.
For more information about enterprise support plans, visit jozu.com/kitops-modelpack-support.