Skip to main content

auto-deploy-usage

Glows.ai Auto Deploy User Guide​

In a typical setup, using GPU deployment services often requires manually creating and releasing instances. For scenarios where GPU usage is sporadic or infrequent, this manual process can be inefficient and cumbersome.

To address this, Glows.ai offers the Auto Deploy feature. With just a one-time setup, the system will automatically manage the lifecycle of GPU instances on your behalf. Once configured, you will receive a fixed service link. When a request is sent to this link, Glows.ai will automatically create an instance based on your configuration, process the request, and return the result. If no new requests are received within a 5-minute window, the system will automatically release the instance—no manual action needed.

This guide will walk you through how to configure and use the Auto Deploy feature using the BreezyVoice WebUI image as an example.

Step 1: Create a New Deployment Task​

  1. Go to the Auto Deploy page and click the New Deploy button in the top-right corner to create a new deployment task.

image-20250527162345042

  1. Enter a task name and description to help identify and manage your deployment later.

image-20250527162731979

  1. Select the required GPU type and runtime environment for your application. You may use a custom-created Snapshot, or choose from one of the system-provided images.

image-20250529100940587

  1. Configure the Port your service will listen on and the Start Command to launch your application. Then click Confirm to finish the setup.   In this example, the service will run on port 8080, and the executable file is api.py located in the BreezyVoice directory. The configuration is as follows:
Port: 8080
Start Command: cd /BreezyVoice && python api.py

image-20250527162828692

  1. Once the deployment task is created, you will receive a dedicated service link along with your configuration details.

image-20250527162928849

  1. To use the service, simply replace your original API endpoint with the Auto Deploy service link. If your service includes a custom route, append the path to the end of the service link. In this example, the deployed API route is /v1/audio/speech. Here's how to make a request:
curl -X POST "https://tw-01.sgw.glows.ai:xxxxxx/v1/audio/speech" \
 -H "Authorization: Bearer sk-template" \
 -H "Content-Type: application/json" \
 --data '{
  "model": "tts-1",
  "voice": "alloy",
  "input": "How about playing basketball after school? The weather looks great today."
}' --output test_speech.wav

image-20250527163237581

  1. After the request is completed, if no additional requests are received within 5 minutes, the instance will be automatically released.   You can monitor the total cost and the current Instance Status of your deployment task from the Auto Deploy dashboard.   The instance status definitions are as follows:
  • Standby: The deployment task is correctly configured, but no instance is currently running.
  • Idle: The instance is either in the process of being created (upon receiving a request) or being released (after the request has been handled).
  • Running: The instance has been successfully created and is currently handling requests. If no new request is received within 5 minutes, the instance will be automatically released.

image-20250527163744664

Contact Us​

If you have any questions or suggestions while using Glows.ai, feel free to reach out to us via email or Line.

Glows.ai Email : support@glows.ai

Line : https://lin.ee/fHcoDgG