mClouds Offers a Golden Mean for AI Experiments: 48 GB of VRAM at a Price That Won’t Make You Flinch

Subscribe Us:

Hosting provider mClouds has decided that artificial intelligence developers deserve a reasonable compromise between modest computational capabilities and ruinous bills. Instead of pushing the excessive and expensive NVIDIA L40S, the company has introduced a configuration featuring two NVIDIA L4 GPUs with 24 GB each within a single virtual machine. This maneuver delivers the coveted 48 GB of video memory, allowing users to tackle tasks for which a single card is already catastrophically insufficient, while upgrading to higher-tier models would blow a hole in the budget.

Hardware Under the Hood

The configuration of the new X16-48GB L4 tariff is as follows:

This balance of resources is aimed at those deploying large language models (LLMs), experimenting with neural network fine-tuning, or handling image generation and video processing.

The base tariff price already includes certain infrastructural conveniences: a connection speed of up to 120 Mbps, L2–L4 DDoS attack protection to maintain service availability, and the option to choose current versions of Windows, CentOS, Debian, or Ubuntu. Backup services can be activated in the control panel, and scaling CPU cores, RAM, or disk space can be performed either at the time of ordering or during operation, should the sudden need for additional power arise.