KServe Monitoring Card

The KServe Status card (kserve_status) provides real-time visibility into KServe model serving infrastructure across your managed clusters.

What It Shows

• InferenceService status — Ready/Not Ready state for each deployed model
• Replica counts — Current vs desired replicas per InferenceService
• Latency metrics — P50/P95 request latency where available via DCGM or Prometheus
• Error rates — Inference request failures per InferenceService
• Framework — Model framework (TensorFlow, PyTorch, ONNX, etc.)

Adding the Card

1. Open Console Studio from the sidebar (the Add more… button)
2. Search for KServe Status
3. Drag it to your dashboard or click Add

Requirements

• KServe operator installed on managed clusters
• ClusterRole with read access to serving.kserve.io resources
• Optional: Prometheus/DCGM metrics for latency data

Demo Mode

When no live KServe installation is detected, the card displays demo data showing a representative set of InferenceServices with varying status states.