# Red Hat AI 3.5 places control, security, and insight at the heart of enterprise AI deployments > Source: > Published: 2026-09-09 14:06:25+00:00 **Red Hat introduces AI 3.5, an update for the entire AI portfolio focusing on control, security, and transparency across hybrid cloud environments. Organizations receive new tools for verified security, GPU control, and efficient deployment of AI agents.** Red Hat launches AI 3.5, a follow-up to the [previously launched AI 3.3](https://itdaily.com/news/datacenter/red-hat-ai-enterprise-v-3-3/), which focuses on providing scalability, security, and operational control. IT and platform teams in particular gain more control over their AI workloads and infrastructure. The update is relevant for companies working with hybrid clouds that need insight into the usage and performance of their AI applications. Red Hat AI 3.5 is available starting today, also as part of Red Hat AI Factory with Nvidia. ## Pre-emptive security and real-time monitoring EvalHub, now generally available within Red Hat AI 3.5, enables companies to test their AI models for security before they go into production. The tool performs risk-based benchmarks and automatically generates reports demonstrating whether models meet compliance requirements. This makes the AI rollout process more transparent and secure. New dashboards provide real-time insight into the health of AI models, GPU consumption, and performance. This allows platform teams to proactively make adjustments and minimize risks. Built-in monitoring also provides usage measurement per user and clear performance dashboards. ## Control over GPUs and agents Companies with shared infrastructure benefit from stronger multi-tenancy. Hosted control planes on top of Red Hat OpenShift Virtualization make it possible to provide a separate control layer for each customer, while the hardware remains centrally managed. GPUs are distributed fairly via fair-share scheduling, and priority-aware processing gives precedence to critical AI tasks. Red Hat AI 3.5 also introduces new capabilities for AI agents. With AutoRAG, corporate data is linked directly to agents, and there is support for multiple languages and contextual lookup. Ready-to-use agent templates accelerate applications such as code review and document processing.