Skip to content

Same agent Smaller bill. The tools and infrastructure to get more work from your AI. Fewer tokens. Less compute. Lower cost.

[Fix]: Token usage and model name extraction in Langchain instrumentation (#512)

Yehia Amer0eceb7a8a093
diff --git a/sdk/python/pyproject.toml b/sdk/python/pyproject.toml
index f1a001b6..8b383fdd 100644
--- a/sdk/python/pyproject.toml
+++ b/sdk/python/pyproject.toml
@@ -1,6 +1,6 @@
[tool.poetry]
name = "openlit"
-version = "1.30.3"
+version = "1.30.4"
description = "OpenTelemetry-native Auto instrumentation library for monitoring LLM Applications and GPUs, facilitating the integration of observability into your GenAI-driven projects"
authors = ["OpenLIT"]
repository = "https://github.com/openlit/openlit/tree/main/openlit/python"
diff --git a/sdk/python/src/openlit/instrumentation/langchain/langchain.py b/sdk/python/src/openlit/instrumentation/langchain/langchain.py
index 1fcd49d0..fcc86439 100644
--- a/sdk/python/src/openlit/instrumentation/langchain/langchain.py
+++ b/sdk/python/src/openlit/instrumentation/langchain/langchain.py
@@ -481,11 +481,12 @@ def chat(gen_ai_endpoint, version, environment, application_name,
response = wrapped(*args, **kwargs)
try:
- input_tokens = response.response_metadata.get("prompt_eval_count", 0)
- output_tokens = response.response_metadata.get("eval_count", 0)
+ token_usage = response.response_metadata.get("token_usage", {})
+ input_tokens = token_usage.get("prompt_tokens", 0)
+ output_tokens = token_usage.get("completion_tokens", 0)
+ model = response.response_metadata.get("model_name", "gpt-4")
prompt = "" if isinstance(args[0], list) else args[0]
- model = getattr(instance, 'model_name', getattr(instance, 'model', 'gpt-4'))
# Calculate cost of the operation
cost = get_chat_model_cost(
@@ -620,11 +621,12 @@ def achat(gen_ai_endpoint, version, environment, application_name,
response = await wrapped(*args, **kwargs)
try:
- input_tokens = response.response_metadata.get("prompt_eval_count", 0)
- output_tokens = response.response_metadata.get("eval_count", 0)
+ token_usage = response.response_metadata.get("token_usage", {})
+ input_tokens = token_usage.get("prompt_tokens", 0)
+ output_tokens = token_usage.get("completion_tokens", 0)
+ model = response.response_metadata.get("model_name", "gpt-4")
prompt = "" if isinstance(args[0], list) else args[0]
- model = getattr(instance, 'model_name', getattr(instance, 'model', 'gpt-4'))
# Calculate cost of the operation
cost = get_chat_model_cost(
model,