AI GPU服务器

租用GPU算力，运行本地LLM、AI代理和高性能AI应用。

Cloud GPU

AWS (EC2)

Wide range of GPU instances in worldwide AWS infrastructure.

B200H200H100A100Trainium

Azure

GPU virtual machines integrated with Microsoft Azure services.

B200H200H100A100T4

CoreWeave

Specialized GPU cloud for enterprise AI and HPC workloads.

GB200$9.51/hB200$8.6/hH200$1.25/hH100$6.16/hA100$2.25/h

DataCrunch (Verda)

GPU cloud with A100/H100 for training and inference of AI models.

B200H200H100A100L40S

FluidStack

Flexible GPU cloud with competitive pricing for AI workloads.

GB200B200H200H100A100

Google Cloud (GCE)

Enterprise GPU instances integrated with Google Cloud ecosystem.

B200H200H100A100L4

Hyperstack

欧洲 GPU 云,提供按需 H100/A100 和按分钟计费。

H200$1.3/hH100$0.7/hA100$0.35/hL40$0.35/hRTX A6000

Jarvis Labs

GPU cloud optimized for AI model training and experiments.

H100$1.49/hA100$0.41/hRTX 6000RTX 5000A6000

Lambda Cloud

Cloud GPU servers optimized for deep learning and AI research.

B200$1.99/hH200$1.29/hH100$0.79/hA100$0.69/hGH200$1.29/h

OVHcloud

欧洲云巨头,在欧盟数据中心提供 GPU 实例。

H100$0.91/hA100$1/hL40S$0.91/hL4$0.6/hV100S$0.88/h

Paperspace (DigitalOcean)

Cloud GPU by DigitalOcean for ML, AI, and data science.

H100$0.76/hA100$0.76/hRTX 5000$0.51/hRTX 4000$0.45/hA6000$0.76/h

RunPod

On-demand and serverless GPU for ML/AI. Docker container support.

B200$2.09/hH200$1.79/hH100$1.59/hA100$1.09/hRTX 5090$0.49/h

Scaleway

法国云服务商,在欧洲数据中心提供 H100 和 L40S 实例。

H100L40SL4RENDER-S

Together AI

GPU cloud for training and inference of AI models with API access.

B200$1.99/hH200$1.99/hH100$1.99/hA100

Vultr

Cloud GPU instances with global availability and simple API.

B200H200H100A100L40S

Peer-to-peer GPU cloud

Salad Cloud

Distributed GPU cloud with the lowest prices on the market.

RTX 5090$0.16/hRTX 4090$0.08/hRTX 3090$0.04/hRTX 3080$0.04/h

TensorDock

Affordable GPU servers with hourly billing.

H100$1.8/hA100$0.35/hRTX 5090RTX 4090$0.12/hA6000

Vast.ai

Peer-to-peer GPU cloud with the lowest prices. Community and datacenter offerings.

H200H100A100RTX 5090RTX 4090

Dedicated GPU Servers

Latitude.sh

Dedicated bare metal GPU servers for demanding AI applications.

H200H100$1.68/hA100L40S

数据更新于:2026/9/7 · 价格仅供参考,基于公开价目表。以服务商官方价格为准。

What can you use GPU servers for?

Run local LLMs (Llama, Mistral, Qwen), train custom AI models, AI agent and assistant inference, image and video generation, scientific computing and simulations.