Nebius (NASDAQ:NBIS) has unveiled version 3.6 of its AI Cloud platform, known as “Aether,” introducing a range of new capabilities focused on infrastructure automation, security, governance and storage performance.
The update is designed to enhance the user experience for developers, machine learning engineers and enterprise customers building and deploying artificial intelligence applications.
New AI Agent Enables Natural-Language Infrastructure Management
A key addition in the latest release is Nebius Echo, an AI-powered assistant integrated directly into the platform’s web console.
The agent allows users to manage cloud infrastructure through natural-language commands, enabling them to ask questions, retrieve information and perform operational tasks without relying on traditional command-line tools or manual workflows.
Nebius said Echo includes safeguards intended to reduce the risk of accidental actions while interacting with infrastructure resources.
The assistant is powered by open-source models hosted on Nebius Token Factory, the company’s inference platform.
Future enhancements are expected to include infrastructure debugging capabilities and support for multi-step infrastructure-as-code deployments.
Expanded Security and Governance Tools
Version 3.6 also introduces several new security and compliance features aimed at enterprise customers.
Among the additions is a Key Management Service (KMS), allowing organizations to manage their own encryption keys while supporting cryptographic erasure capabilities.
The platform now also supports Workload Identity Federation, enabling credential-free authentication and reducing reliance on static credentials.
To improve financial oversight, Nebius has launched a new Budgets feature that allows teams to establish spending targets and receive alerts when predefined usage thresholds are reached across projects.
Storage Enhancements Deliver Higher Performance
The update includes a number of improvements designed to increase storage efficiency and performance.
Nebius has introduced local SSD storage options for GPU-based servers and launched an Intelligent Object Storage Class that automatically transfers archived data to lower-cost storage tiers without imposing request or egress charges.
The company also reported a 30% increase in read bandwidth for single-threaded Object Storage workloads.
Additional upgrades to the shared filesystem significantly improve performance, delivering three times more input/output operations per second (IOPS) for 4K file operations and up to 100 times more IOPS for metadata-intensive workloads.
Nebius said the platform has now been validated to support storage clusters as large as 100 petabytes.
Builder Program Targets Developers and AI Engineers
Alongside the platform update, Nebius introduced the Nebius Builder Program in an early preview phase.
The initiative is designed to support developers and machine learning engineers by providing credits across the company’s cloud, inference and learning platforms.
Participants will also gain access to engineering office hours and additional technical resources intended to accelerate development projects.
Certification Program Debuts
The company simultaneously launched Nebius Certifications through Nebius Academy.
The first credential available is an associate-level AI Cloud Operations certification, aimed at validating foundational cloud and infrastructure skills on the Nebius platform.
Nebius said additional certification pathways are expected to be introduced in the coming weeks as the company expands its training and education offerings.
Company Expands AI Cloud Ecosystem
The release of AI Cloud 3.6 reflects Nebius’ ongoing effort to broaden its ecosystem and strengthen its position within the rapidly growing AI infrastructure market.
By combining automation tools, enhanced security features, improved storage performance and new training initiatives, the company is seeking to make its platform more attractive to enterprises and developers building next-generation AI applications.