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
Products
Start from your terminal or from your agent
Two products on the same GPUs. Pick the one that fits how you work.
Sandboxes
Scalable programmatic microVMs for agentic workloads that requires GPUs.
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
The same GPUs for a fraction of the price
An A100 80GB is $1.09/hr on Thunder Compute and $3.43 on AWS. No contracts, no minimums.
Price per GPU-hour
LOVED BY DEVELOPERS
“This pricing is insane.”
“Can't get enough of this UX.”
Real quotes from ML engineers building on Thunder Compute
Enterprise
Thunder Compute
on your own GPUs
Bring Thunder Compute to your data center. Get more work out of the GPUs you already own.
One pool for every team
Teams draw GPUs from a shared pool instead of each owning cards that sit idle.
Your hardware
Runs on your GPU machines. Set up and managed with the thunder CLI.
node-01
- train
- train
- train
- train
- train
- train
- eval
- idle
node-02
- infer
- infer
- infer
- train
- train
- train
- train
- notebook
node-03
- eval
- eval
- train
- train
- infer
- infer
- notebook
- idle
node-04
- notebook
- notebook
- notebook
- eval
- eval
- infer
- infer
- train
ml-research · inference · evals · notebooks, one pool30/32 busy
Research
From the lab
Research and engineering from the team building Thunder Compute.