V100S

The NVIDIA V100S is built on the Volta architecture with 32 GB of HBM2 VRAM. As an enhanced variant of the original V100, it delivers improved clock speeds and memory bandwidth for datacenter AI workloads, maintaining scalability in multi-GPU configurations.

With 32 GB of VRAM, the V100S supports Large Language Models (LLM) and fine-tuning of moderately sized models. While newer architectures have surpassed its raw throughput, it remains a cost-effective option for organizations running established training pipelines and inference workloads on existing Volta-based infrastructure.