Instances
A GPU instance you connect to and keep. Snapshot it when you are done.
~ zsh
$ tnr create --gpu a100
✓ Instance 0 created
$ tnr connect 0
✓ Connected to instance 0
ubuntu@instance-0:~$ nvidia-smi -L
GPU 0: NVIDIA A100 80GB PCIe
The agent-first GPU cloud
GPU instances for developers. Full Ubuntu sandboxes for agents running ML experiments, CUDA code, and other GPU workloads.
Products
Instances for developers, GPU sandboxes for agents
Instances for hands-on development and sandboxes for RL environments and evals at scale
Sandboxes
Run RL tasks, evals, and CI jobs that need a GPU in isolated Ubuntu VMs. Scale out in parallel and pay per second for the compute each task needs.
A full computer for your agent
Run the tools, dependencies, and services your task needs in its own Ubuntu VM.
agent.py
import thunder_sandbox as thunder
sandbox = thunder.Sandbox.create(
cpu=4, memory=32, storage=50,
gpu_type=thunder.GPUType.H100, gpu_count=1,
timeout=900)
sandbox.wait_until_ready()
process = sandbox.exec("nvidia-smi")
Pricing
Stop wasting runway on GPUs
Scale GPU sandboxes with your workload, pay per second, and shut them down when the work is done.
price per
Loved by developers
“Can’t get enough of this UX.”
“We are saving so much on GPU Sandboxes.”
Real quotes from ML engineers and RL researchers building on Thunder Compute
Research
From the lab
Research and engineering from the team building Thunder Compute.
Start building today
Pick the form factor that suits your workload
Instances
Long-lived GPU servers for manual development
From $0.35/hrBilled by the minuteSandboxes
A full Ubuntu computer for your agent, with GPU access, VM isolation, and per-second billing.
From $1.95/GPU-hrBilled per second