LiteLLM Instrumentation

Installation

$ pip install respan-instrumentation-litellm

Classes

LiteLLMInstrumentor

Respan instrumentation plugin that registers a LiteLLM CustomLogger callback and emits canonical chat spans through respan-tracing.

1 from respan_instrumentation_litellm import LiteLLMInstrumentor

Parameter	Type	Default	Description
`include_content`	`bool`	`True`	Capture request messages and assistant response content on chat spans.

Methods

Method	Description
`activate()`	Register the Respan callback in `litellm.callbacks`.
`deactivate()`	Remove the callback registered by this instrumentor.

RespanLiteLLMCallback

Low-level callback class used by LiteLLMInstrumentor. Most applications should initialize it through Respan(instrumentations=[LiteLLMInstrumentor()]).

Usage

1 import litellm
2 from respan import Respan
3 from respan_instrumentation_litellm import LiteLLMInstrumentor
4 
5 respan = Respan(
6     api_key="your-api-key",
7     app_name="litellm-service",
8     instrumentations=[LiteLLMInstrumentor()],
9 )
10 
11 response = litellm.completion(
12     model="gpt-4o-mini",
13     messages=[{"role": "user", "content": "Tell me a joke"}],
14 )
15 print(response.choices[0].message.content)
16 respan.flush()

Async usage

1 import litellm
2 from respan import Respan
3 from respan_instrumentation_litellm import LiteLLMInstrumentor
4 
5 respan = Respan(instrumentations=[LiteLLMInstrumentor()])
6 
7 response = await litellm.acompletion(
8     model="gpt-4o-mini",
9     messages=[{"role": "user", "content": "Tell me a joke"}],
10 )
11 respan.flush()

LiteLLM integration guide