English 
HiReso has launched a dedicated plan for its new NVIDIA B200 powered multi node high end GPU cluster under the GPUSOROBAN lineup. The big shift here is that the company is moving beyond single server rentals and tying multiple GPU servers together over a high speed network so users can push heavier loads and run large parallel jobs. This basically opens the door for serious generative AI work including LLM training and other compute hungry experiments that were hard to run on isolated servers.
The cluster uses a Slurm based job scheduling setup so teams can squeeze more efficiency out of the hardware and control both time and cost more tightly. The dedicated plan gives customers guaranteed access to the latest B200 GPUs and multi node configurations at a time when GPU availability is still a headache across the industry.
こちらもお読みください： ABB Schweiz、業務シフトのためにAzureとGen AIを採用
ハイレゾ is also prepping a points based system where users buy credits and consume GPU time in one minute chunks with one-year validity and commercial use allowed. That adds flexibility for teams that do not want long reservations but still need access to high end compute when workloads spike. The company says more details on that system are coming soon.
日本語