(All-cores)
(Single-core)
Specifications
Server Metadata
Vendor ID | gcp |
Server ID | 1740006 |
Name | ct5lp-hightpu-8t |
Description | 224 vCPUs, 384 GB RAM, 8 Google TPUs |
Family | ct5lp |
Hw Virt | - |
Status | active |
Observed At | 2026-10-25T21:12:47.930280 |
Availability
| REGION / ID | SPOT | ONDEMAND |
|---|---|---|
Council Bluffs (US) / us-central1 | 2.8746 USD/h | 9.6 USD/h |
The Dalles (US) / us-west1 | 2.8746 USD/h | 9.6 USD/h |
Moncks Corner (US) / us-east1 | 2.8746 USD/h | 9.6 USD/h |
Ashburn (US) / us-east4 | 3.9096 USD/h | 9.6 USD/h |
Las Vegas (US) / us-west4 | 3.944 USD/h | 9.6 USD/h |
Columbus (US) / us-east5 | 4.56 USD/h | 9.6 USD/h |
Montréal (CA) / northamerica-northeast1 | 0.9514 USD/h | 10.5706 USD/h |
St. Ghislain (BE) / europe-west1 | 4.2282 USD/h | 10.5706 USD/h |
Hamina (FI) / europe-north1 | 5.2853 USD/h | 10.5706 USD/h |
Changhua County (TW) / asia-east1 | 2.2825 USD/h | 11.1158 USD/h |
Frankfurt (DE) / europe-west3 | 1.0195 USD/h | 11.328 USD/h |
Dallas (US) / us-south1 | 6.3326 USD/h | 11.328 USD/h |
Tokyo (JP) / asia-northeast1 | 1.1096 USD/h | 12.3283 USD/h |
Eemshaven (NL) / europe-west4 | 4.7241 USD/h | 12.48 USD/h |
Jurong West (SG) / asia-southeast1 | 6.24 USD/h | 12.48 USD/h |
Santiago (CL) / southamerica-west1 | 1.2355 USD/h | 13.728 USD/h |
Processor
vCPUs | 224 |
CPU Allocation | Dedicated |
CPU Architecture | x86_64 |
System Resources and Accelerators
| MEMORY | |
|---|---|
Memory Amount | 384 GiB |
| ACCELERATORS | |
|---|---|
Accelerator Count | 8 |
Accelerator Memory Min | 16 GiB |
Accelerator Memory Total | 128 GiB |
Accelerator Type | TPU |
Accelerator Manufacturer | Google |
Accelerator Family | TPU |
Accelerator Model | v5e |
Accelerators |
|
| STORAGE | |
|---|---|
Storage Size | 0 GB |
Storages |
| NETWORK | |
|---|---|
Inbound Traffic | 0 GB/month |
Outbound Traffic | 0 GB/month |
IPv4 | 0 |
Server Description
A high-density tensor processing instance featuring eight integrated accelerators and dedicated compute cores for parallelized machine learning workloads.
Google Cloud Platform ct5lp-hightpu-8t is a specialized, TPU-accelerated server instance designed for machine learning and deep learning workloads. Built on the x86_64 architecture, it features 224 dedicated vCPUs and 384.0 GB of system memory, providing 1.71 GB of RAM per core. The core acceleration is driven by eight integrated Google TPU v5e chips with a combined 128 GB of VRAM. This instance does not include local storage, requiring external storage configurations for data pipelines. The high density of dedicated vCPUs combined with specialized tensor processing units makes this instance highly efficient for large-scale model training, transformer inference, and neural network computations. It is optimized for workloads requiring parallelized matrix operations rather than general-purpose compute or high local storage capacity.
Economics
Alternatives
Servers of the Same Family
| INSTANCE | vCPUs | MEMORY | Accelerators |
|---|---|---|---|
| ct5lp-hightpu-1t | 24 | 48 GiB | 1 |
| ct5lp-hightpu-4t | 112 | 192 GiB | 4 |
Similar Servers
| INSTANCE | VENDOR | vCPUs | MEMORY | Accelerators |
|---|