Products

One network,
every way to compute.

GPU compute, pre-built environments, and dedicated VMs on a distributed network built for resilience. Deploy anything from inference endpoints to CI/CD pipelines.

Get started free View pricing

deploy3 products

GPU Computeready

Pre-Built Environmentsready

Dedicated Virtual Machinesready

global edge networkon demand

GPU Compute

Rent the machine.

Access GPUs on demand across our distributed network. Deploy inference and training workloads with automatic failover — when nodes fail, your workloads don't.

H100, H200, A100, RTX 4090, RTX 3090 and more

Automatic failover and load balancing

NVLink and NVSwitch for multi-GPU workloads

Starting at $0.19/hr

Browse GPUs

gpu_inventoryon demand

H100

80 GB

iad1

$2.30/hr

A100

80 GB

sfo3

$0.80/hr

RTX 4090

24 GB

ord1

$0.37/hr

Pre-Built Environments

Skip the setup.

Deploy pre-configured VMs built for specific workflows — CI/CD runners, AI agents, and more. Every environment ships with all dependencies pre-installed, ready in minutes.

Self-hosted GitHub Actions runners with GPU access

AI agent environments with browser automation

Zero setup — all dependencies pre-installed

Isolated VMs with VFIO GPU passthrough

View Environments

ci_runneronline

jobs:
  train:
    runs-on: [self-hosted, gpu, openrelay]
    steps:
      - run: nvidia-smi
      - run: pytest tests/

runner-iad-041× H100no queue

Dedicated Virtual Machines

Full root access.

Spin up GPU-accelerated VMs with full root access. Bring your own container image or start from a base Ubuntu VM with NVIDIA drivers and Docker pre-installed.

Full root SSH access

Docker and NVIDIA drivers pre-installed

Persistent storage across reboots

Custom container image support

Deploy a VM

ssh root@vm-7f3aconnected

$ nvidia-smi --query-gpu=name,memory.total --format=csv,noheader
NVIDIA H100 80GB HBM3, 81559 MiB

$ docker run --gpus all my-model:latest
[ ready ] serving on :8000

Pricing