PRODUCT

Servers for AI.
100% API.

Create GPU and CPU servers, then scale them. 100× cheaper, 100× faster, 100× better than AWS. CUDA, root, NVMe, unlimited bandwidth. One REST API.

Create a serverAPI docs

HOW IT WORKS

Create. Scale. Infer.

01
$ curl -X POST api.rawhq.io/signup
{"email":"you@lab.ai"}
✓ token: raw_live_…

Get an API token

One POST. Bearer token. That's the entire account. No AWS IAM maze.

02
$ curl -X POST api.rawhq.io/deploy
{"type":"raw-gpu-44","region":"eu"}
✓ GPU live (3s) · CUDA + root
$304/mo · 100× vs AWS

Create a server

Pick a GPU or CPU type and a region. Live in 3 seconds. Public IP. Full root.

03
$ curl -X POST api.rawhq.io/deploy
# same call. N more GPUs.
✓ fleet: 8 GPUs · vLLM ready
✓ scale down: DELETE /servers/:id

Scale

The same API creates the next box. Resize, rebuild, destroy. Per-second billing.

100% API

The API is the product

Create servers, scale a fleet, destroy when the job ends. CLI and dashboard wrap the same endpoints.

API

REST with 24 endpoints. Create servers, attach volumes, inject SSH keys, pull metrics. Bearer token. JSON. Built so agents and CI can run your fleet.

API reference →

CLI

npx rawhq deploy — create, scale, SSH, destroy. Same API, nicer shell. JSON output for scripts.

npm install -g rawhq

Dashboard

When you want a UI: metrics, graphs, networking, volumes, backups, snapshots, terminal. Same servers the API created.

Open dashboard →

FEATURES

Everything included

3-second create

POST /deploy to a running GPU with a public IP. Faster than AWS describes the instance.

Full root + CUDA

Your GPU, your weights. Install vLLM, Ollama, PyTorch. No SageMaker lock-in.

NVMe storage

All servers use NVMe SSDs. Fast reads, fast writes, no spinning disks.

5 global regions

Germany, Ireland, US East, US West, Singapore. Deploy close to your users.

SSH key management

Add keys via CLI, dashboard, or import from GitHub. Manage per-server access.

Live metrics

CPU, RAM, disk I/O, and network graphs. Per-core utilization. Auto-refreshing.

Web terminal

VNC console access directly in the browser. No SSH client needed.

Snapshots and backups

Point-in-time snapshots on demand. Automated daily backups with 7-day retention.

Volumes

Attach block storage that persists independently. 10 GB to 10 TB. $0.07/GB/month.

Flat pricing

One price per server. No egress fees. No hidden charges. No surprises on the bill.

Unlimited bandwidth

Included with every server. No metering, no overage charges.

100% API

Create, scale, resize, destroy. 24 REST endpoints. Bearer token. Built for agents and CI.

INFRASTRUCTURE

Dedicated hardware for AI

NVIDIA GPUs and AMD EPYC across five regions. CUDA, NVMe, unlimited bandwidth. 100× cheaper than AWS GPU.

Regions

🇩🇪Frankfurt, Germanyeu
🇮🇪Dublin, Irelandeu-fi
🇺🇸Ashburn, Virginiaus
🇺🇸Hillsboro, Oregonus-west
🇸🇬Singaporesg

Server types

2–16 vCPU4–32 GB · from $9/mo
32–48 vCPU128–192 GB · from $680/mo
GPU20–96 GB VRAM · from $304/mo

OS images

Ubuntu 24.04Ubuntu 22.04Debian 12Fedora 40CentOS Stream 9Rocky 9

GPU SERVERS

GPUs you create over API

Dedicated NVIDIA for inference, fine-tunes, and training. Full CUDA. 100× cheaper than AWS GPU. GDPR EU.

Inference

20 GB VRAM

raw-gpu-44
$304/mo
CPU14 cores
RAM64 GB
NVMe3.8 TB
ModelsLlama 8B, Whisper, SD
Training

96 GB VRAM

raw-gpu-131
$1,510/mo
CPU24 cores
RAM256 GB
NVMe1.9 TB
ModelsLlama 70B, custom training
Max

96 GB VRAM

raw-gpu-131p
$2,914/mo
CPU24 cores
RAM768 GB
NVMe15 TB
ModelsLargest configs
View all GPU plans →

SERVER MANAGEMENT

Full control from one place

Every server gets a dedicated management page with 14 tabs — modeled after the professional cloud console. Metrics, graphs, networking, volumes, backups, snapshots, rescale, rebuild, power management, rescue system, and a web terminal.

OverviewServer status, IP addresses, specs, root password, quick actions
GraphsCPU, network, and disk I/O charts with configurable time ranges
NetworkingPublic and private IPs, reverse DNS, network configuration
FirewallsInbound traffic rules and port management
VolumesCreate, attach, and manage persistent block storage
BackupsEnable automated daily backups with one click
SnapshotsCreate point-in-time server images on demand
RescaleUpgrade or downgrade server specs without losing data
RebuildReinstall the operating system from scratch
PowerPower on, power off, and power cycle controls
RescueBoot into a recovery environment for troubleshooting
DeletePermanently destroy the server and all associated data
TerminalBrowser-based VNC console with full keyboard support
AICopy-paste prompts for Claude or ChatGPT to manage your server

Create your first AI server

100% API. 3 seconds. 100× cheaper than AWS.

$ curl -X POST https://api.rawhq.io/deploy
Get startedView pricing