The platform is available now, upgrading existing Nebius AI Studio users and supporting over 60 open-source models across text, code, and vision.

The Microsoft agreement underscores strong hyperscaler interest in partnering with emerging neocloud providers.

Early adopters include Prosus, Higgsfield AI, and Hugging Face, with reported benefits like up to 26x cost reductions and scalable, production-ready inference.

Co-founder notes the product is designed to optimize inference pipelines, deliver production-ready systems, and enable scaling with minimal overhead.

Nebius has been rapidly expanding in the US, Europe, and Israel, including opening one of Israel’s first public clusters using Nvidia’s latest chips.

The launch follows Nebius’s international expansion and its Yandex spin-off, reinforcing its growing global presence and Nvidia-backed capabilities.

Nebius has a multibillion-dollar deal with Microsoft worth up to $19.4 billion to provide computing capacity, signaling strong validation from major tech players.

Nebius Group launches Token Factory, a platform that gives developers access to open-source AI models and runs inference workloads on Nebius’s high-performance compute clusters.

Token Factory supports models such as OpenAI GPT-oss, Meta Llama, and China's DeepSeek, enabling users to mix models for their needs while leveraging Nebius’s compute power for deployment.

Analyst sentiment remains bullish on NBIS, with a Strong Buy consensus and an average target around $149.50, suggesting about 35% upside.

The move aims to expand Nebius’s enterprise reach and provide flexibility beyond closed AI ecosystems, potentially reducing customization costs for clients.

Token Factory is built on Nebius AI Cloud 3.0 “Aether,” emphasizing enterprise-grade security, monitoring, and performance validated by benchmarks such as MLPerf Inference.