MLServices

Query Prometheus for service-level metrics

get/api/v1/mlservices/{name}/metrics

Request

  • Base URL: / (relative, and the document was published on a file host, not at the API’s origin)
  • URL: /api/v1/mlservices/{name}/metrics
  • Auth: none declared

Path parameters

namestring required

Query parameters

metric'request_rate' | 'latency' | 'error_rate' | 'cpu_util' | 'mem_util' | 'gpu_util' required
rangestring required

ISO 8601 duration (5m, 15m, 1h, 6h, 24h).

stepstring
percentile'p50' | 'p95' | 'p99'

Only for metric=latency.

Response

Metric series.

metricstring required

Metric name (e.g. qps, latency).

percentilestring

Percentile the series represents (e.g. p95).

rangestring required

Query time range (e.g. 1h).

stepstring

Sampling step between points (e.g. 1m).

unitstring

Value unit (e.g. ms, req/s).

Example response

{
  "metric": "qps",
  "percentile": "p95",
  "range": "1h",
  "series": [
    {
      "timestamp": "2026-06-28T09:00:00Z",
      "value": 12.5
    },
    {
      "timestamp": "2026-06-28T09:30:00Z",
      "value": 18.3
    },
    {
      "timestamp": "2026-06-28T09:25:00Z",
      "value": 15.1
    }
  ],
  "step": "1m",
  "unit": "req/s"
}

Changes

No recorded changes to this operation across all 17 revisions of this API.