Nebius Introduces Token Factory for Scalable AI Inference Deployment

Nebius has unveiled the Nebius Token Factory, designed for vertical AI companies and digital enterprises to scale and optimize open-source and custom models. Built on its AI infrastructure, the platform brings together model lifecycle processes with high-performance inference and fine-grained access management.

The platform supports major open models like NVIDIA Nemotron and DeepSeek, offering the capability to host custom models. Nebius Token Factory aims to eliminate bottlenecks from closed models, promoting innovation with open-source models. It features sub-second latency, autoscaling, and 99.9% uptime for extensive AI solutions.

Prosus, a prominent lifestyle and e-commerce brand, reported significant cost reductions. Meanwhile, Higgsfield AI and Hugging Face leverage Nebius for scalable and reliable AI applications.

The Nebius Token Factory, part of the Nebius AI Cloud 3.0 “Aether,” ensures security and performance, facilitating fast scaling of AI projects through its robust infrastructure.