High-performance software framework to schedule AI workload to geo-distributed GPUs and deliver unprecedented throughput on heterogeneous and commodity compute resources to beat the performance of high-end GPUs.

Enable AI training, fine-tuning and inference on a wide spectrum of GPUs ranging from commodity to high-end ones with limited compute and memory capacity.

Built-in support and optimization to run major open-source LLMs on a highly elastic GPU network. Allow customization for new LLMs to achieve the state-of-the-art results.