List observations

Lists gen_ai spans as LLM observations — each an LLM call with its model, token counts, cost and latency projected from gen_ai.* attributes, newest first, over the query window.

GET /v1/o11y/llm/observations

O11y → · All API references → · Get started →

Address	`https://api.hanzo.ai/v1/o11y/llm/observations`
Method	`GET`
Operation	`ListLLMObservations`
Auth	`Authorization: Bearer $HANZO_API_KEY`

Lists gen_ai spans as LLM observations — each an LLM call with its model, token counts, cost and latency projected from gen_ai.* attributes, newest first, over the query window.

Callers need the viewer role; the runtime's own gate enforces it, and scopes the read to the caller's validated tenant.

Request

9 fields.

Field	In	Type	Required	Description
`start`	query	integer	—	Start is the start of the window as a unix-millisecond epoch.
`end`	query	integer	—	End is the end of the window as a unix-millisecond epoch.
`traceId`	query	string	—	TraceID narrows the view to one trace.
`sessionId`	query	string	—	SessionID narrows the view to one conversation.
`userId`	query	string	—	UserID narrows the view to one end user.
`name`	query	string	—	Name narrows the view to observations of one name.
`model`	query	string	—	Model narrows the view to one model.
`offset`	query	integer	—	Offset is how many rows to skip, for paging.
`limit`	query	integer	—	Limit caps how many rows come back.

Response

Status	Body	Meaning
`200`	o11y.O11yLLMObservationsOut	ok
`202`	Approval	held for approval
`default`	problem-details	refused

200 body — 22 fields.

Field	In	Type	Always	Description
`data`	body	o11y.O11yLLMObservationsPage	—
`data.items`	body	o11y.O11yLLMObservation[]	—	Items are the observations, newest first.
`data.items[].completionTokens`	body	integer (int64)	—	CompletionTokens is the output token count.
`data.items[].id`	body	string	—	ID is the observation's id (the span id).
`data.items[].latencyMs`	body	number (double)	—	LatencyMs is how long it took, in milliseconds.
`data.items[].model`	body	string	—	Model is the model that served it.
`data.items[].name`	body	string	—	Name is the observation's name.
`data.items[].parentObservationId`	body	string	—	ParentID is the parent observation, when the span has one.
`data.items[].promptTokens`	body	integer (int64)	—	PromptTokens is the input token count.
`data.items[].provider`	body	string	—	Provider is the model's provider.
`data.items[].serviceName`	body	string	—	ServiceName is the app that emitted it.
`data.items[].sessionId`	body	string	—	SessionID is the conversation the observation belongs to.
`data.items[].startTime`	body	string (date-time)	—	StartTime is when the observation started.
`data.items[].statusCode`	body	string	—	StatusCode is the observation's status, e.g. OK, ERROR.
`data.items[].totalCost`	body	number (double)	—	TotalCost is the observation's cost.
`data.items[].totalTokens`	body	integer (int64)	—	TotalTokens is the sum of prompt and completion tokens.
`data.items[].traceId`	body	string	—	TraceID is the trace the observation belongs to.
`data.items[].type`	body	string	—	Type is the observation kind, e.g. chat, embeddings, tool.
`data.items[].userId`	body	string	—	UserID is the end user the observation is attributed to.
`data.limit`	body	integer (int64)	—	Limit is the page cap the read ran with.
`data.offset`	body	integer (int64)	—	Offset is the row offset this page started at.
`status`	body	string	—	Status is "success".

Failure carries the platform error shape — see Errors.

Examples

CLI

SDK

HTTP

MCP

hanzo o11y llm observations

.ts

.py

.go

.rs

.java

import { Configuration, O11yApi } from 'hanzoai';

const api = new O11yApi(new Configuration({ accessToken: process.env.HANZO_API_KEY }));
const { data } = await api.listLLMObservations();

from hanzoai.cloud import ApiClient, Configuration
from hanzoai.cloud.api import O11yApi

client = ApiClient(Configuration(access_token=os.environ["HANZO_API_KEY"]))
result = O11yApi(client).list_llm_observations()

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

resp, _, err := client.O11yAPI.ListLLMObservations(context.Background()).Execute()
if err != nil {
	return err
}

use hanzo_client::apis::{configuration::Configuration, o11y_api};

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

let result = o11y_api::list_llm_observations(&cfg, Default::default()).await?;

import ai.hanzo.cloud.ApiClient;
import ai.hanzo.cloud.api.O11yApi;

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

var result = new O11yApi(client).listLLMObservations();

curl https://api.hanzo.ai/v1/o11y/llm/observations \
  -H "Authorization: Bearer $HANZO_API_KEY"

Tool o11y, op ListLLMObservations — 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": "o11y",
         "arguments": {
           "op": "ListLLMObservations",
           "input": {}
         }
       }
     }'

O11y API · All Hanzo APIs · Interactive reference

Was this page useful?

PreviousList annotation NextList llm pricing rules