AI research and development at Nebius
The Nebius AI R&D team pursues application-oriented AI research, democratizing progress by publicly sharing our findings and assets such as data, models and code.
Our blog
Projects
Kvax
Our open-source library offering fast and efficient attention operations for the JAX framework. Built with FlashAttention-2 algorithms implemented in the Triton language, it is optimized for high-performance attention computation with document masks and supports context parallelism.
SWE-rebench
We present SWE-rebench, a new benchmark for evaluating agentic LLMs on a continuously updated and decontaminated set of real-world software engineering tasks mined from real GitHub repositories.
