GPU Finder

Main features include global public cloud GPU inventory statistics and data retrieval. It supports multi-dimensional filtering by GPU specification (such as H100, H200, L4, L40S, A100, Tesla series, NVIDIA G series, etc.), GPU count (ranging from 1/12 to 16), memory size (8GB to 2048GB), vCPU cores (2 to 256), deployment region (covering dozens of global nodes including Beijing, Hong Kong, Frankfurt, Tokyo, Silicon Valley, etc.), and cloud platform (Alibaba Cloud, Amazon Cloud, Azure, Google Cloud, Huawei Cloud, Tencent Cloud, Volc Engine).

Pricing information provides costs for both prepaid and postpaid models, detailing the total hourly and monthly costs as well as the hourly and monthly amortized costs per GPU card, displayed in CNY by default with an option to switch to USD. The core advantage is aggregating real-time inventory and pricing of public cloud GPUs across the internet, helping users quickly compare the costs and availability of different instance specifications across various cloud providers, greatly improving the efficiency of compute procurement and architecture decision-making. Typical use cases include AI model training and inference deployment, cloud server GPU selection, and enterprise IT cost evaluation and optimization.