Red Hat AI 3.4 has been updated to simplify the development and deployment of intelligent agent workflows across hybrid cloud environments.
The company announced these updates on June 22, 2026, to bridge the gap between experimental projects and production-level control. Consequently, organizations can now scale their artificial intelligence models across their entire infrastructure.
Features of Red Hat AI 3.4
This platform update introduces a Model-as-a-Service framework to provide a unified interface for accessing certified models. Meanwhile, system administrators can track consumption and apply policies using high-performance distributed inference.
Specifically, this inference is powered by vLLM and llm-d technologies to maintain efficiency across diverse environments.
Managing Agentic Workflows and Security
New AgentOps tools manage autonomous agents from development to production through integrated tracking and observability. Furthermore, the platform treats prompt management as a core data asset alongside a dedicated evaluation center.
To secure these workflows, the system uses automated adversarial scanning to detect vulnerabilities and prompt injection risks. Notably, these security measures integrate with cybersecurity protocols to protect enterprise data.
Executive Perspectives on Infrastructure
Joe Fernandes, vice president and general manager of AI Business at Red Hat, stated that the era of agentic AI represents a new phase.
“The era of agentic AI represents a new phase in our platform’s evolution, moving from running traditional applications to enabling intelligent, autonomous systems.”
Joe Fernandes, Vice President and General Manager of AI Business at Red Hat
In addition, John Fanelli, vice president of enterprise software at NVIDIA, stated that autonomous agents require infrastructure control.
Hardware Support and Availability
The updated Red Hat AI 3.4 platform provides immediate support for NVIDIA Blackwell GPUs and AMD MI325X processors. Moreover, it extends its architecture to run on managed clouds, including IBM Cloud, to ensure operational consistency.
According to the company, the Red Hat AI 3.4 release is expected to become available during June 2026.
Transitioning to Token Providers
Many organizations now need to transition from being token consumers to becoming token providers to manage costs. However, the gap between developers and infrastructure administrators remains a major challenge for widespread adoption.
This platform addresses this challenge by providing a foundation for deploying inference and autonomous agents at scale. Consequently, businesses can maintain transparency and governance to meet compliance and risk management standards.
