Skip to main content
GET
/
api
/
v2
/
external
/
inference
/
replica
/
{replica_id}
/
machine-metrics
Get replica machine metrics
curl --request GET \
  --url https://api.example.com/api/v2/external/inference/replica/{replica_id}/machine-metrics
{
  "replicaId": "<string>",
  "hostname": "<string>",
  "timeRange": {
    "start": "<string>",
    "end": "<string>"
  },
  "gpuUtilizationPercent": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuMemoryUtilizationPercent": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuTemperatureCelsius": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuPowerWatt": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuPowerLimitWatt": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuClockSmMhz": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuClockMemMhz": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuPcieThroughputRxBytesPerSec": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "gpuPcieThroughputTxBytesPerSec": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "systemRamTotalBytes": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "systemRamUsedBytes": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ],
  "systemCpuUsagePercent": [
    {
      "timestamp": "<string>",
      "value": 123
    }
  ]
}

Path Parameters

replica_id
string
required

Query Parameters

start
string | null

ISO 8601 start timestamp (defaults to replica.created_at)

end
string | null

ISO 8601 end timestamp (defaults to replica.last_health_check if terminated, otherwise now)

step
string
default:15s

Query resolution (e.g., '5s', '15s', '1m')

Response

Successful Response

Resource-agnostic machine-level GPU+system metrics response.

replicaId
string
required
hostname
string
required
timeRange
TimeRange · object
required

Time range for the metrics query.

gpuUtilizationPercent
MetricDataPoint · object[]
required
gpuMemoryUtilizationPercent
MetricDataPoint · object[]
required
gpuTemperatureCelsius
MetricDataPoint · object[]
required
gpuPowerWatt
MetricDataPoint · object[]
required
gpuPowerLimitWatt
MetricDataPoint · object[]
required
gpuClockSmMhz
MetricDataPoint · object[]
required
gpuClockMemMhz
MetricDataPoint · object[]
required
gpuPcieThroughputRxBytesPerSec
MetricDataPoint · object[]
required
gpuPcieThroughputTxBytesPerSec
MetricDataPoint · object[]
required
systemRamTotalBytes
MetricDataPoint · object[]
required
systemRamUsedBytes
MetricDataPoint · object[]
required
systemCpuUsagePercent
MetricDataPoint · object[]
required