More Granular Observability for Vercel Sandbox

Vercel has released detailed metrics features that provide deep insight into Vercel Sandbox's resource consumption. Now, through the Observability tab in the dashboard, you can monitor compute and networking usage in real time.

The key monitoring items are as follows:

• Active CPU and CPU Usage: Measures the time your code actually uses the CPU, with I/O wait time excluded from billing.
• Provisioned Memory: Measures allocated memory multiplied by execution time (GB-hours).
• Data Transfer: Measures total data sent and received, including package downloads and API calls.
• Running Sandboxes and Sessions: Check the number of currently running Sandboxes and the start/end times of sessions.

All metrics can be grouped by Sandbox Name and Sandbox Session ID, allowing you to track usage of individual instances. Additionally, the ability to query and visualize metrics via the Vercel CLI is also provided.

This update is available at the team and project level, and is designed to match the actual billing method. This allows you to track AI agent workloads that create Sandboxes at scale, optimize settings according to resource usage, and prevent unexpected costs in advance.