Install & authenticate
pip install lowcostgpu
export LOWCOSTGPU_API_KEY=pg_live_… # console → API keys
lowcostgpu whoami
# account ok · balance $41.22 · 2 instances runningThe key can also live in ~/.config/lowcostgpu/config.toml — the CLI reads the environment first, then the file. Create keys in the console (key docs).
Find capacity
# the price sheet, in your terminal
lowcostgpu gpus --sort price
# machines matching a shape
lowcostgpu offers --gpu rtx-4090 --num 2 --region eu-west-1
# ID GPU × REGION vCPU RAM DISK $/HR
# m-8b02c1f3 RTX 4090 2 eu-west-1 16 64GB 1TB 0.708Launch & manage
lowcostgpu launch --gpu rtx-4090 --template comfyui --disk 60
# ✓ instance i-b81f02aa running (28.4s)
# ✓ https://i-b81f02aa.lowcostgpu.com:8188 (comfyui)
lowcostgpu list # your fleet, states, burn/hr
lowcostgpu logs i-b81f02aa -f # stream container logs
lowcostgpu stop i-b81f02aa # GPU billing ends this second
lowcostgpu start i-b81f02aa
lowcostgpu destroy i-b81f02aa --yesVolumes & snapshots
lowcostgpu volume create --name models --size 500 --region eu-west-1
lowcostgpu launch --gpu l40s --template vllm --volume models:/models:ro
lowcostgpu snapshot create i-b81f02aa --name goldenScripting patterns
Every command takes --json for machine-readable output, and exit codes follow convention (0 ok, 1 API error, 2 usage):
# deploy, wait, run, destroy — a disposable GPU in a Makefile
ID=$(lowcostgpu launch --gpu rtx-4090 --template pytorch --json | jq -r .id)
lowcostgpu wait "" --state running
lowcostgpu exec "" -- python train.py
lowcostgpu destroy "" --yesShell completion
lowcostgpu completion bash >> ~/.bashrc # zsh & fish tooBuilding in Python instead? The SDK page shows the same flows as typed calls.