Dashboard Metrics Reference

Reference for dashboard cards, chart metrics, and interpretation hints

Overview

This reference summarizes key Dashboard metrics and how to interpret them during routine operations.

Summary Cards

Total Requests

• Represents: recent aggregate request activity.
• Displayed with: formatted total, trend indicator, and percentage delta.
• Use for: rapid demand-change detection.

Models

• Represents: total available models.
• Displayed with: cloud and local split.
• Use for: inventory and hosting-mix awareness.

Endpoints

• Represents: endpoint footprint and running subset.
• Use for: immediate service readiness checks.

Clusters

• Represents: cluster availability posture.
• Use for: capacity/risk awareness before traffic peaks.

Projects

• Represents: active project scope and membership.
• Use for: understanding potential demand contributors.

Analytical Charts

API Calls

• Dimension: project.
• Metric: request volume by selected window.
• Interpretation: identifies load concentration.

Latency

• Dimension: project.
• Metric: response time trend.
• Interpretation: detects reliability degradation.

Throughput

• Dimension: project.
• Metric: processing rate.
• Interpretation: reveals efficiency or saturation changes.

Model Usage

• Dimension: model (top-ranked set).
• Metric: usage frequency/volume.
• Interpretation: tracks production model popularity.

Accuracy

• Dimension: top models across benchmarks.
• Metric: comparative benchmark scores.
• Interpretation: supports quality-aware routing decisions.

Token Metrics

• Dimension: top models.
• Metric: input and output token totals.
• Interpretation: cost and prompt-behavior proxy.

Time Windows

UI option	Typical use
Last 24 Hrs	Incident triage and short-lived anomalies
Last 7 Days	Team operational review and weekly trends
Last 30 Days	Capacity planning and strategic optimization