Cirrascale

CuratedActive
cirrascale.comUSUnited States, San DiegoSince 2010

San Diego AI cloud with NVIDIA, AMD, Qualcomm and Tenstorrent systems on term pricing

GPU cloudInference
Visit website PricingContacts
From
—
GPU from
$3.85/GPU-hr
Locations
0
countries
Data centers
0
sites
Networks (ASN)
0
ASNs
Plans & prices
3
plans

Overview

Cirrascale Cloud Services rents dedicated multi-accelerator servers (NVIDIA B200/H200, AMD Instinct, Qualcomm Cloud AI and Tenstorrent) on monthly to annual terms with flat "no surprises" billing, and runs an inference platform including hosted Ai2 model endpoints.

Plans & prices

GPU instances3
InstanceConfigper GPU-hrTotal /hrSpotLocationsSource
8× AMD MI300X server, monthly term ($22,499/mo) — per-GPU-hour equivalent
1× Instinct MI300X$3.85$3.85—
 Verified
3 days ago
8× NVIDIA H200 server, monthly term ($26,499/mo) — per-GPU-hour equivalent
1× H200$4.54$4.54—
 Verified
3 days ago
8× NVIDIA B200 server, monthly term ($34,999/mo) — per-GPU-hour equivalent
1× B200$5.99$5.99—
 Verified
3 days ago

GPU models

B200RTX PRO 6000 BlackwellH200H100 SXMA100 40GBRTX A6000RTX A5000RTX A4000Instinct MI300XInstinct MI250

Similar providers

DigitalOceanDeveloper-focused cloud known for simple Droplet VMs, now also selling GPU DropletsUSVultrIndependent global cloud with VMs, bare metal and NVIDIA/AMD GPUs in 30+ regionsUSRunPodDeveloper GPU cloud with per-second Pods, serverless endpoints and community hostsUSCoreWeaveLargest independent AI neocloud, renting NVIDIA GB300, B200 and H200 clustersUSLambdaGPU cloud for AI training with on-demand B200/H100 instances and 1-Click ClustersUSCrusoeVertically integrated AI cloud building its own power-first data center campusesUSMassed ComputeUS GPU cloud renting B200, H200 NVL, H100 and workstation GPUs by the hourUSTogether AIOpen-model inference API plus self-serve and reserved NVIDIA GPU clustersUS