DEVELOPER TOOLS

GPU Cloud from Your Terminal

Search thousands of GPUs, deploy Docker containers, and manage instances — all from the command line. No UI, no waitlist, no contracts.

Get API Key View Docs

vast-terminal

Install in seconds

pip install vastai
vastai set api-key YOUR_API_KEY

Requires Python 3.9+. Also available via pipx.

How it works

Step 1

Search the Marketplace

Filter by GPU model, VRAM, price, reliability, location, and 20+ fields. Sort by performance-per-dollar.

$ vastai search offers 'gpu_name=H100_SXM num_gpus>=4 reliability>0.99' -o 'dph' --limit 5

  ID       GPU         Num   VRAM    $/hr    DLPerf   Location
  847291   H100_SXM    4     320GB   $4.12   98.7     US-East
  851003   H100_SXM    4     320GB   $4.28   97.2     EU-West
  849177   H100_SXM    8     640GB   $7.84   99.1     US-West

Step 2

Deploy Your Container

Launch any Docker image on bare-metal GPUs. Pass environment variables, expose ports, mount volumes. SSH access included.

$ vastai create instance 847291 \
    --image nvcr.io/nvidia/pytorch:24.01-py3 \
    --disk 64 --ssh --direct

  {"success": true, "new_contract": 9841205}

Step 3

Monitor & Manage

Check status, stream logs, execute commands, and destroy instances when done. Script your entire workflow.

$ vastai show instances

  ID        GPU          Status    Image                              Cost/hr
  9841205   4x H100_SXM  running   nvcr.io/nvidia/pytorch:24.01-py3   $4.12

$ vastai logs 9841205 --tail 5

  [14:02:45] Model loaded. 70B params, 4-bit quantized.
  [14:02:46] Server ready on :8080

$ vastai destroy instance 9841205

20+ Filter Fields

GPU model, VRAM, price, reliability, location, bandwidth, compute capability, and more.

Any Docker Image

Public or private registries. PyTorch, TensorFlow, vLLM, custom images — anything.

Spot Pricing

Bid on interruptible instances. RTX 4090 from $0.14/hr.

SSH & Jupyter

Direct SSH access or launch Jupyter notebooks with --jupyter.

File Transfer

vastai copy moves data between instances or to/from local.

Scripting-First

JSON output with --raw. Pipe to jq. Automate everything.

GPUs from $0.14/hr (spot) and $0.50/hr (on-demand)

Thousands of GPUs available across 30+ data centers worldwide.

View Pricing

Deploy your first GPU in 60 seconds

No contracts, no minimums. Get your API key and start building.

Get API Key Read Full Docs