Implementation:BerriAI Litellm OpenMeter Logger

Attribute	Value
Sources	`litellm/integrations/openmeter.py`
Domains	Logging, Metering, Usage Tracking, Integrations
Last Updated	2026-02-15 16:00 GMT

Overview

The OpenMeterLogger sends LLM usage events (token counts, cost, model) to OpenMeter for metering and billing purposes, using the CloudEvents specification.

Description

OpenMeterLogger extends CustomLogger to post CloudEvents-formatted usage data to the OpenMeter API (/api/v1/events). Each event includes the model name, token usage (prompt, completion, total), cost, a subject identifier (derived from the user parameter or user_api_key_user_id metadata), and a timestamp. The event type defaults to "litellm_tokens" and can be overridden with the OPENMETER_EVENT_TYPE environment variable. Both sync and async logging methods are supported. The logger requires a user identifier -- either passed directly in the completion call or available from the API key metadata.

Usage

Import and register OpenMeterLogger when you need to meter LLM usage for billing. Requires OPENMETER_API_KEY environment variable.

Code Reference

Source Location

litellm/integrations/openmeter.py

Signature

class OpenMeterLogger(CustomLogger):
    def __init__(self) -> None

Import

from litellm.integrations.openmeter import OpenMeterLogger

I/O Contract

Inputs

Parameter	Type	Required	Description
(none)	--	--	Constructor takes no arguments. Validates environment variables on init.

Environment Variables:

Variable	Required	Description
`OPENMETER_API_KEY`	Yes	OpenMeter API key for authentication.
`OPENMETER_API_ENDPOINT`	No	API base URL. Defaults to `"https://openmeter.cloud"`.
`OPENMETER_EVENT_TYPE`	No	CloudEvent type string. Defaults to `"litellm_tokens"`.

Key Methods

Method	Returns	Description
`log_success_event(kwargs, response_obj, start_time, end_time)`	`None`	Sync logging of usage event to OpenMeter.
`async_log_success_event(kwargs, response_obj, start_time, end_time)`	`None`	Async logging of usage event to OpenMeter.

Outputs

Output	Type	Description
Side effect	HTTP POST	Sends a CloudEvents JSON payload to `{endpoint}/api/v1/events`.

CloudEvent payload structure:

{
    "specversion": "1.0",
    "type": "litellm_tokens",
    "id": "<call_id>",
    "time": "<ISO 8601 timestamp>",
    "subject": "<user_id>",
    "source": "litellm-proxy",
    "data": {
        "model": "<model_name>",
        "cost": 0.001,
        "prompt_tokens": 10,
        "completion_tokens": 20,
        "total_tokens": 30
    }
}

Usage Examples

import litellm

# Register the callback
litellm.success_callback = ["openmeter"]

response = litellm.completion(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}],
    user="user-123",  # Required: identifies the subject for metering
)

# Direct instantiation
from litellm.integrations.openmeter import OpenMeterLogger

logger = OpenMeterLogger()

Related Pages

BerriAI_Litellm_PostHog_Logger - another analytics/metering integration
BerriAI_Litellm_PromptLayer_Logger - another simple logging integration
BerriAI_Litellm_Braintrust_Logger - another observability integration

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment