Nebius Group (NASDAQ: NBIS) announced the release of Nebius AI Cloud 3.5, which introduces serverless capabilities and additional GPU hardware options to its cloud platform.

The serverless features allow developers to launch workloads without configuring infrastructure, with the platform handling configuration and runtime management. The service is available in public preview.

The company expanded its GPU offerings to include NVIDIA RTX PRO 6000 Blackwell Server Edition, targeting workloads such as AI inference, industrial robotics, physical AI simulations, visual computing, and drug discovery. This hardware is available immediately.

Version 3.5, called "Aether," includes a Data Transfer Service designed to simplify data migration and replication between external S3-compatible storage systems and Nebius cloud regions. The update also modifies the configuration setup for Managed Soperator, the company's Slurm-on-Kubernetes solution, providing additional options for creating Slurm clusters.

The release includes updates to Managed Kubernetes observability for cluster-level control and a redesigned AI application marketplace. Additional changes include improved user administration with role-based permissions and new public APIs for billing data export.

All features in the Aether 3.5 release are available on Nebius's global AI Cloud infrastructure, according to the company's announcement.