Implementation:BerriAI Litellm PromptLayer Logger

Attribute	Value
Sources	`litellm/integrations/prompt_layer.py`
Domains	Logging, Prompt Management, Integrations
Last Updated	2026-02-15 16:00 GMT

Overview

The PromptLayerLogger sends LLM completion events to the PromptLayer platform for request tracking and metadata logging.

Description

PromptLayerLogger is a simple logger that posts LLM call data to PromptLayer's REST API (/rest/track-request). On each success event, it sends the model name, messages, optional parameters, response object, timestamps, API key, and any PromptLayer-specific tags (extracted from metadata["pl_tags"]). If PromptLayer returns a request_id, the logger also posts additional metadata to the /rest/track-metadata endpoint. The response object is converted from Pydantic models to dictionaries for JSON serialization. This is one of the simpler integrations in LiteLLM -- it does not extend CustomLogger or CustomBatchLogger, instead providing a standalone log_event() method.

Usage

Import and register PromptLayerLogger when you want to track LLM requests in PromptLayer. Requires PROMPTLAYER_API_KEY environment variable.

Code Reference

Source Location

litellm/integrations/prompt_layer.py

Signature

class PromptLayerLogger:
    def __init__(self)

Import

from litellm.integrations.prompt_layer import PromptLayerLogger

I/O Contract

Inputs

Parameter	Type	Required	Description
(none)	--	--	Constructor takes no arguments. Reads `PROMPTLAYER_API_KEY` from environment.

Key Methods

Method	Returns	Description
`log_event(kwargs, response_obj, start_time, end_time, print_verbose)`	`None`	Logs the completion event to PromptLayer and optionally posts metadata.

Method parameters:

Parameter	Type	Description
`kwargs`	`dict`	Request kwargs containing `model`, `messages`, `optional_params`, and `litellm_params`.
`response_obj`	`Any`	The LLM response object (ModelResponse or dict).
`start_time`	`datetime`	Request start time.
`end_time`	`datetime`	Request end time.
`print_verbose`	`Callable`	Verbose logging function.

Outputs

Output	Type	Description
Side effect	HTTP POST	Posts to `https://api.promptlayer.com/rest/track-request` and optionally `/rest/track-metadata`.

Usage Examples

import litellm

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

response = litellm.completion(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}],
    metadata={
        "pl_tags": ["production", "experiment-v2"],
        "custom_key": "custom_value",
    },
)

# Direct instantiation
from litellm.integrations.prompt_layer import PromptLayerLogger

logger = PromptLayerLogger()
logger.log_event(
    kwargs={"model": "gpt-4", "messages": [...], "optional_params": {}, "litellm_params": {"metadata": {}}},
    response_obj=response,
    start_time=start_time,
    end_time=end_time,
    print_verbose=print,
)

Related Pages

Implementation:BerriAI_Litellm_OpenMeter_Logger - another simple logging integration
Implementation:BerriAI_Litellm_Braintrust_Logger - a more feature-rich observability integration
Implementation:BerriAI_Litellm_Humanloop_Integration - another prompt management integration

Page Connections

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

Principle

Implementation

Heuristic

Environment