List samples

GET /v1/compute/fleet/samples

Compute → · All API references → · Get started →


Address	`https://api.hanzo.ai/v1/compute/fleet/samples`
Method	`GET`
Operation	`listFleetSamples`
Auth	`Authorization: Bearer $HANZO_API_KEY`

Returns the caller org's utilization series, oldest first.

A rejected narrower is a 400 carrying its own reason (the vocabulary is ours and safe to echo); a warehouse failure is logged and answered 503 "unavailable", because a chart that silently reads "no load" when the truth is "we cannot tell" is worse than one that says so. An ABSENT warehouse is different again: it returns an empty series, which renders honestly as "no samples yet".

Request

3 fields.

Field	In	Type	Required	Description
`unit`	query	string	—	Unit selects one compute unit's series by its source-local id.
`source`	query	string	—	Source selects one plane: "agent", "byo" or "visor".
`range`	query	string	—	Range is the lookback window (e.g.

Response

Status	Body	Meaning
`200`	compute.sampleList	ok
`default`	problem-details	refused

200 body — 37 fields.

Field	In	Type	Always	Description
`samples`	body	compute.sampleView[]	—	Samples are the readings, OLDEST first — the order a chart plots.
`samples[].at`	body	string	—	At is when the reading was MEASURED, RFC 3339 in UTC — the x-axis a chart plots against.
`samples[].costCents`	body	integer (int64)	—	CostCents is what this unit resold for over the hour the reading falls in, in whole US cents.
`samples[].cpuTemp`	body	number (double)	—	CPUTemp is the CPU package temperature in °C.
`samples[].cpuUtil`	body	number (double)	—	CPUUtil is the busy fraction of all cores over the sample window, 0..1.
`samples[].cpus`	body	integer (int64)	—	CPUs is logical cores. The static capability rides every row on purpose: a chart can size load against cores without joining a registry whose row may since…
`samples[].decode`	body	number (double)	—	Decode is generation throughput over the sample window, in tokens per second.
`samples[].diskRead`	body	number (double)	—	DiskRead is the block-device read rate in bytes per second.
`samples[].diskTotal`	body	integer (int64)	—	DiskTotal is the size of those filesystems, in BYTES.
`samples[].diskUsed`	body	integer (int64)	—	DiskUsed is the space used across local block filesystems, in BYTES.
`samples[].diskWrite`	body	number (double)	—	DiskWrite is the block-device write rate in bytes per second.
`samples[].gpuMemTotal`	body	integer (int64)	—	GPUMemTotal is dedicated VRAM in BYTES; absent when the GPU shares system memory (unified).
`samples[].gpuMemUsed`	body	integer (int64)	—	GPUMemUsed is the memory held by GPU processes, in BYTES.
`samples[].gpuModel`	body	string	—	GPUModel names the representative accelerator ("GB10"); GPUs carries how many.
`samples[].gpuPower`	body	number (double)	—	GPUPower is the total GPU power draw in watts.
`samples[].gpuPowerLimit`	body	number (double)	—	GPUPowerLimit is the total enforced GPU power limit in watts; absent when the driver reports none.
`samples[].gpuTemp`	body	number (double)	—	GPUTemp is the hottest GPU's temperature in °C.
`samples[].gpuUtil`	body	number (double)	—	GPUUtil is aggregate accelerator utilization as a FRACTION of 1 — 0.42 is 42% busy.
`samples[].gpus`	body	integer (int64)	—	GPUs is how many accelerators the reading covers.
`samples[].host`	body	string	—	Host is the hostname the unit reported at the time of the reading.
`samples[].kind`	body	string	—	Kind is what the measured unit is: laptop, cloud, gpu, cluster, machine or worker.
`samples[].kvCache`	body	number (double)	—	KVCache is the KV-cache occupancy, 0..1.
`samples[].load1`	body	number (double)	—	Load1 is the 1-minute load average — runnable processes, not a percentage.
`samples[].load5`	body	number (double)	—	Load5 is the 5-minute load average, the same units as Load1.
`samples[].load15`	body	number (double)	—	Load15 is the 15-minute load average, the same units as Load1.
`samples[].memFree`	body	integer (int64)	—	MemFree is host memory available, in BYTES, as reported rather than derived.
`samples[].memUsed`	body	integer (int64)	—	MemUsed is host memory in use, in BYTES.
`samples[].memory`	body	integer (int64)	—	Memory is total system RAM in BYTES at the time of the reading.
`samples[].model`	body	string	—	Model is the model id(s) served on the unit, comma-separated, at most 128 bytes.
`samples[].netRx`	body	number (double)	—	NetRx is the receive rate of physical interfaces in bytes per second.
`samples[].netTx`	body	number (double)	—	NetTx is the transmit rate of physical interfaces in bytes per second.
`samples[].prefill`	body	number (double)	—	Prefill is prompt (prefill) throughput over the sample window, in tokens per second.
`samples[].running`	body	integer (int64)	—	Running is the number of requests being served at the reading.
`samples[].source`	body	string	—	Source is the plane that reported the reading: "agent", "byo" or "visor" — the same vocabulary the board's rows carry, and what ?source= narrows on.
`samples[].ttft`	body	number (double)	—	TTFT is the mean time to first token, in seconds, of requests that started in the window.
`samples[].unit`	body	string	—	Unit is the source's own id for the measured unit.
`samples[].waiting`	body	integer (int64)	—	Waiting is the number of requests queued at the reading.

Failure carries the platform error shape — see Errors.

Examples

hanzo compute fleet samples get

import { Configuration, ComputeApi } from 'hanzoai';

const api = new ComputeApi(new Configuration({ accessToken: process.env.HANZO_API_KEY }));
const { data } = await api.listFleetSamples();

from hanzoai.cloud import ApiClient, Configuration
from hanzoai.cloud.api import ComputeApi

client = ApiClient(Configuration(access_token=os.environ["HANZO_API_KEY"]))
result = ComputeApi(client).list_fleet_samples()

cfg := hanzoai.NewConfiguration()
cfg.AddDefaultHeader("Authorization", "Bearer "+os.Getenv("HANZO_API_KEY"))
client := hanzoai.NewAPIClient(cfg)

resp, _, err := client.ComputeAPI.ListFleetSamples(context.Background()).Execute()
if err != nil {
	return err
}

use hanzo_client::apis::{configuration::Configuration, compute_api};

let mut cfg = Configuration::new();
cfg.bearer_access_token = std::env::var("HANZO_API_KEY").ok();

let result = compute_api::list_fleet_samples(&cfg, Default::default()).await?;

import ai.hanzo.cloud.ApiClient;
import ai.hanzo.cloud.api.ComputeApi;

ApiClient client = new ApiClient();
client.setBearerToken(System.getenv("HANZO_API_KEY"));

var result = new ComputeApi(client).listFleetSamples();

curl https://api.hanzo.ai/v1/compute/fleet/samples \
  -H "Authorization: Bearer $HANZO_API_KEY"

Tool visor, op listFleetSamples — POST the JSON-RPC envelope to https://api.hanzo.ai/v1/mcp.

curl -X POST https://api.hanzo.ai/v1/mcp \
  -H "Authorization: Bearer $HANZO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
       "jsonrpc": "2.0",
       "id": 1,
       "method": "tools/call",
       "params": {
         "name": "visor",
         "arguments": {
           "op": "listFleetSamples",
           "input": {}
         }
       }
     }'

Compute API · All Hanzo APIs · Interactive reference

Request

Response

Examples

On this page