OpenVINO™ Telemetry

2025.2.0 · active · verified Sat Apr 11

OpenVINO™ Telemetry is a Python 3 library (current version 2025.2.0) designed to collect anonymous usage and performance data from OpenVINO™ toolkit components. It sends events, comprising a category, action, and label, to Google Analytics to aid in debugging and further development. The library is actively maintained as part of the broader OpenVINO™ toolkit, with releases typically aligning with OpenVINO™ versions.

Warnings

gotcha Telemetry data collection is opt-out by default, but critically, it *requires explicit user consent* during the OpenVINO toolkit installation. If this consent is not provided, or if the user has previously opted out via `opt_in_out --opt_out`, no data will be transmitted. This can lead to silently failing data collection if the consent status is not actively managed or verified.
Fix: Developers integrating `openvino-telemetry` should be aware of the consent mechanism. Ensure that users grant consent or, for testing/development, explicitly enable telemetry using `opt_in_out --opt_in` if data is expected to be sent.
gotcha When providing detailed information in the `label` parameter of `send_event`, it is highly recommended to combine all related key-value pairs into a single, JSON-like string (e.g., `'{key1:value1, key2:value2}'`). Sending separate events for related data points can lead to duplicated metrics and make it difficult to correlate information effectively in analytics.
Fix: Always construct a single stringified dictionary (or a similar structured string) for the `label` argument to `send_event` when you want to track connected data points for a single event.
gotcha Anonymous telemetry data collected by OpenVINO™ Telemetry is stored in Google Analytics with a maximum retention period of 14 months. Raw data older than this threshold is periodically deleted.
Fix: Factor in the 14-month data retention policy when designing analytics dashboards or performing historical data analysis based on telemetry from `openvino-telemetry`.

Install

pip install openvino-telemetry Install latest version

Imports

telemetry
```
import openvino_telemetry as telemetry
```
The primary way to interact with the library is by importing it, often aliased as 'telemetry', and then calling its functions like `send_event`.
send_event
```
from openvino_telemetry import send_event
```
Directly import the `send_event` function for convenience if it's the only needed component.

Quickstart

This quickstart demonstrates how to import the `openvino_telemetry` library and send anonymous usage events. Events consist of a `category` (tool name), `action` (metric/function), and `label` (detailed data, often a JSON-like string). Telemetry collection is opt-out by default and requires user consent, which can also be managed via the `opt_in_out` command line utility.

import openvino_telemetry as telemetry
import os

# Telemetry is collected by default but requires user consent during OpenVINO installation.
# If the control file doesn't exist or indicates 'no' consent, no data is transmitted.
# For explicit control (e.g., in a test environment):
# os.system("opt_in_out --opt_out") # To disable
# os.system("opt_in_out --opt_in")  # To enable

# Send a simple event with a category, action, and label.
# The label can be a simple string or a JSON-like string for structured data.
telemetry.send_event(
    category="my_application", 
    action="start_up", 
    label="{version:1.0.0, os:" + os.name + "}"
)

# Example of sending an event related to a specific OpenVINO tool, as per documentation guidelines.
# Ensure all related data is in the 'label' as a single stringified dictionary for correlation.
telemetry.send_event(
    category="mo", 
    action="conversion_results", 
    label="{framework:pytorch, model:yolov8, status:success, duration_sec:120}"
)

print("Telemetry events sent (if consent is granted).")

view raw JSON →