Aggregate Analytics

While App Analytics drills into a single app, the aggregate analytics dashboard answers the higher-level question: how is everything you have deployed doing right now? It opens automatically at the top of the Apps page and summarizes capacity, traffic, and per-app activity across all of your apps in one place. This is especially useful once you are running tens of apps — whether they are stages of one pipeline or independent endpoints — and need to spot problems without clicking into each one.

​

What the Overview Shows

The dashboard at the top of the Apps page combines several components that aggregate across every app you own:

• GPU and CPU distribution — how your active runners are split across machine types (H100, H200, A100, L40, B200, RTX 6000, and CPU classes), so you can see where your capacity is concentrated.
• Top apps — your highest-volume apps over the selected window (for example, the last 6 hours) by total requests.
• Requests by status code — aggregate request traffic broken down by success, user errors, and server errors.
• Request traffic — requests per second across all apps combined.
• Requests by app — a stacked chart showing which apps are contributing the most requests to your combined usage.

​

Organizing and Scanning Many Apps

Alongside the charts, the Apps page gives you tools to manage a large fleet of apps at a glance:

• List view — a compact row-based layout that makes it practical to work with tens of apps. Each row surfaces key metrics (runners, error rate, latency) plus a per-app request breakdown by status code over time, so you can quickly scan for apps with a high runner count or an elevated error rate.
• Tags — label apps by modality, team, project, or any scheme that fits your workflow (for example, tag an upscaler app as upscaler or a streaming app as real-time), each with its own color.
• Filtering — filter the listing by tag to instantly narrow down to a group of apps, then drill into any specific app that needs a closer look.

​

From Overview to Root Cause

The aggregate view is the starting point of the observability workflow. Use it to notice something worth investigating — an app with a spike in errors, or a machine type consuming more capacity than expected — then drill into that app’s App Analytics, Requests, Runner Analytics, or Logs to find the root cause.