Create Serverless Endpoint

Learn how to create and configure a serverless endpoint for your AI workloads.

What is an Endpoint?

An endpoint is a deployment unit that:

• Runs your worker container image
• Processes tasks from a dedicated queue
• Auto-scales based on demand

Create via Console

Step 1: Navigate to Endpoints

1. Go to wavespeed.ai/serverless
2. Click Endpoints in the left sidebar
3. Click Create Endpoint

Step 2: Configure Basic Settings

Step 3: Configure Scaling

Step 4: Create

Click Create Endpoint and wait for deployment.

Using Private Images

If your image is in a private registry, you need to add credentials first.

Add Registry Credentials

1. Go to Settings in the left sidebar
2. Under Registry Credentials, click Add Credential
3. Enter:

4. Click Save

Use Credentials in Endpoint

When creating an endpoint with a private image, select the credential from the dropdown.

Endpoint Status

Best Practices

Naming Conventions

• Use descriptive names: flux-inference, llm-chat-prod
• Include environment: my-model-dev, my-model-prod

Scaling Configuration

• Development: Min 0, Max 1 (cost-effective)
• Production: Min 1, Max 10 (always available)
• Batch Processing: Min 0, Max 20 (handle bursts)

Resource Allocation

• Start with the smallest GPU that fits your model
• Monitor GPU memory usage and adjust if needed
• Use multi-GPU only if single GPU is insufficient

Managing Endpoints

View Endpoint Details

Click an endpoint name to see:

• Current status and replica count
• Task statistics
• Configuration details

Update Endpoint

1. Click the endpoint name
2. Click Edit
3. Modify settings
4. Click Save

Note: Some changes require redeployment.

Delete Endpoint

1. Click the endpoint name
2. Click Delete
3. Confirm deletion

Deleting an endpoint:

• Stops all workers
• Cancels pending tasks
• Removes the endpoint configuration

Related Pages

• Build Worker — Write your handler code
• GPU Pricing — Choose the right GPU
• API Reference — Submit tasks via API