VoltageGPU CLI

Confidential AI Infrastructure from the command line.

voltagegpu-cli lets you deploy and manage Confidential Compute GPU pods on VoltageGPU — hardware-attested Intel TDX environments with AES memory encryption and Protected PCIe, built for regulated workloads that cannot trust the host.

Install

pip install voltagegpu-cli

Or from source:

git clone https://github.com/voltagegpu/voltagegpu-cli.git
cd voltagegpu-cli
pip install -e .

Configure

Get an API key from voltagegpu.com and set it:

export VOLT_API_KEY="your_api_key_here"

Or create ~/.volt/config.ini:

[api]
api_key = your_api_key_here

Quick start

# Browse available Confidential Compute GPU tiers
volt cc inventory

# Register your SSH public key
volt ssh-keys add --name laptop --file ~/.ssh/id_ed25519.pub

# Deploy a hardware-attested pod
volt cc deploy --name my-pod --resource h200-small --image ubuntu:22.04

# List running pods
volt pods list

# Print the SSH command
volt pods ssh <pod_id>

# Stop or delete
volt pods stop <pod_id>
volt pods delete <pod_id>

Commands

Confidential Compute

# List available GPU tiers with live pricing and availability.
volt cc inventory

# Option A — deploy from scratch with an explicit image + resource.
volt cc deploy --name NAME \
               --resource TIER \
               --image ubuntu:22.04 \
               [--ssh-key <uid>]

# Option B — deploy from a curated template (image + resource come
# from the template manifest; you can still override either).
volt cc deploy --name NAME --template <tpl-uid>

Templates

# List curated templates (Jupyter, DeepSeek, GLM, Kimi, Qwen3, …).
volt templates list [--type RENTAL|SERVERLESS|ALL] [--json]

# Show the full manifest for a single template.
volt templates get <tpl-uid>

Machines

# List Confidential GPU machine tiers with live pricing.
volt machines list [--json]

Pods

volt pods list [--json]
volt pods get <pod_id>
volt pods ssh <pod_id>
volt pods stop <pod_id>
volt pods delete <pod_id> [--yes]

SSH keys

volt ssh-keys list [--json]
volt ssh-keys add --name LABEL (--file PATH | --key "ssh-ed25519 AAAA…")
volt ssh-keys delete <key_id> [--yes]

Account

volt account balance
volt account info
volt config               # show current SDK config (masked)

All list commands support --json for machine-readable output.

Python SDK

from volt import VoltageGPUClient

with VoltageGPUClient() as client:
    # Browse inventory
    for item in client.list_confidential_inventory():
        print(item["display_name"], item["final_price_per_hour"])

    # Browse templates (Jupyter, DeepSeek-R1, GLM-4.6, Kimi, Qwen3-Coder…)
    for tpl in client.list_templates():
        print(tpl.uid, tpl.name, tpl.resource_name)

    # Deploy a pod — either from an explicit image/resource…
    result = client.create_confidential_pod(
        name="training-run",
        resource_name="h200-small",
        image="ubuntu:22.04",
    )
    # …or from a curated template uid.
    result = client.create_confidential_pod(
        name="deepseek-infer",
        template_uid="tpl-5npuuq70m1uo",  # DeepSeek-R1 (8x H200)
    )
    pod = result["pod"]
    print(f"Deployed {pod['targonUid']} — {pod['ssh_command']}")

    # List, stop, delete
    for pod in client.list_pods():
        print(pod.name, pod.status, pod.hourly_price)

    client.stop_pod(pod.id)
    client.delete_pod(pod.id)

Custom configuration

from volt.sdk import Config, VoltageGPUClient

client = VoltageGPUClient(
    config=Config(
        api_key="your_api_key",
        base_url="https://api.voltagegpu.com/api",
    )
)

Environment variables

Variable	Description	Default
VOLT_API_KEY	Your VoltageGPU API key	—
VOLT_BASE_URL	API base URL	https://api.voltagegpu.com/api

Examples

Stop every running pod

volt pods list --json | \
  jq -r '.[] | select(.status == "RUNNING") | .id' | \
  xargs -I {} volt pods stop {}

Pick the cheapest H200 tier

volt cc inventory --json | \
  jq 'map(select(.display_name | test("H200"))) | sort_by(.final_price_per_hour)[0]'

Support

• Website: https://voltagegpu.com
• Docs: https://docs.voltagegpu.com
• Email: support@voltagegpu.com

License

MIT — see LICENSE.