How to beginner · 3 min read

How to launch a RunPod GPU pod

Q: How to launch a RunPod GPU pod

Use the runpod Python package to launch a GPU pod by setting runpod.api_key from your environment and creating an Endpoint instance with your pod ID. Then call endpoint.run_sync() with your input to run jobs on the GPU pod.

Quick answer

Use the runpod Python package to launch a GPU pod by setting runpod.api_key from your environment and creating an Endpoint instance with your pod ID. Then call endpoint.run_sync() with your input to run jobs on the GPU pod.

PREREQUISITES

Python 3.8+
RunPod API key
pip install runpod

Setup

Install the official runpod Python package and set your API key as an environment variable for secure authentication.

bash

pip install runpod

output

Collecting runpod
  Downloading runpod-1.0.0-py3-none-any.whl (10 kB)
Installing collected packages: runpod
Successfully installed runpod-1.0.0

Step by step

This example shows how to launch a RunPod GPU pod synchronously using the pod's endpoint ID. Replace YOUR_ENDPOINT_ID with your actual pod endpoint ID.

python

import os
import runpod

# Set your RunPod API key from environment
runpod.api_key = os.environ["RUNPOD_API_KEY"]

# Create an Endpoint instance with your GPU pod endpoint ID
endpoint = runpod.Endpoint("YOUR_ENDPOINT_ID")

# Run a job synchronously with input prompt
result = endpoint.run_sync({"input": {"prompt": "Hello from RunPod GPU pod!"}})

print("Output:", result["output"])

output

Output: {'response': 'Hello from RunPod GPU pod! Your request was processed successfully.'}

Common variations

You can run jobs asynchronously using endpoint.run_async() and await the result. Also, you can customize the input dictionary to match your pod's expected schema. For streaming or advanced usage, refer to the RunPod SDK docs.

python

import asyncio
import os
import runpod

async def main():
    runpod.api_key = os.environ["RUNPOD_API_KEY"]
    endpoint = runpod.Endpoint("YOUR_ENDPOINT_ID")
    result = await endpoint.run_async({"input": {"prompt": "Async call to RunPod GPU pod"}})
    print("Async output:", result["output"])

asyncio.run(main())

output

Async output: {'response': 'Async call to RunPod GPU pod processed successfully.'}

Troubleshooting

If you get an authentication error, verify your RUNPOD_API_KEY environment variable is set correctly.
If the pod ID is invalid, confirm you copied the correct endpoint ID from your RunPod dashboard.
For timeout errors, check your network connection and pod status on the RunPod console.

✅

Key Takeaways

Set runpod.api_key from environment variable RUNPOD_API_KEY before using the SDK.
Use runpod.Endpoint("YOUR_ENDPOINT_ID") to target your GPU pod.
Call endpoint.run_sync() or endpoint.run_async() to execute jobs on the pod.
Replace input dictionary keys to match your pod's expected input schema.
Check RunPod dashboard for pod status and correct endpoint IDs.

Verified 2026-04

Verify ↗