Wrappers

wrappers

This module provides convenient tracing wrappers for popular libraries.

FUNCTION	DESCRIPTION
wrap_anthropic	Patch the Anthropic client to make it traceable.
wrap_gemini	Patch the Google Gen AI client to make it traceable.
wrap_openai	Patch the OpenAI client to make it traceable.

wrap_anthropic

wrap_anthropic(client: C, *, tracing_extra: TracingExtra | None = None) -> C

Patch the Anthropic client to make it traceable.

PARAMETER	DESCRIPTION
client	The client to patch. TYPE: C
tracing_extra	Extra tracing information. TYPE: TracingExtra | None DEFAULT: None

RETURNS	DESCRIPTION
C	The patched client.

Example

import anthropic
from langsmith import wrappers

client = wrappers.wrap_anthropic(anthropic.Anthropic())

# Use Anthropic client same as you normally would:
system = "You are a helpful assistant."
messages = [
    {
        "role": "user",
        "content": "What physics breakthroughs do you predict will happen by 2300?",
    }
]
completion = client.messages.create(
    model="claude-3-5-sonnet-latest",
    messages=messages,
    max_tokens=1000,
    system=system,
)
print(completion.content)

# You can also use the streaming context manager:
with client.messages.stream(
    model="claude-3-5-sonnet-latest",
    messages=messages,
    max_tokens=1000,
    system=system,
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)
    message = stream.get_final_message()

wrap_gemini

wrap_gemini(
    client: C,
    *,
    tracing_extra: TracingExtra | None = None,
    chat_name: str = "ChatGoogleGenerativeAI",
) -> C

Patch the Google Gen AI client to make it traceable.

Warning

BETA: This wrapper is in beta.

Supports

• generate_content and generate_content_stream methods
• Sync and async clients
• Streaming and non-streaming responses
• Tool/function calling with proper UI rendering
• Multimodal inputs (text + images)
• Image generation with inline_data support
• Token usage tracking including reasoning tokens

PARAMETER	DESCRIPTION
client	The Google Gen AI client to patch. TYPE: C
tracing_extra	Extra tracing information. TYPE: TracingExtra | None DEFAULT: None
chat_name	The run name for the chat endpoint. TYPE: str DEFAULT: 'ChatGoogleGenerativeAI'

RETURNS	DESCRIPTION
C	The patched client.

Example

from google import genai
from google.genai import types
from langsmith import wrappers

# Use Google Gen AI client same as you normally would.
client = wrappers.wrap_gemini(genai.Client(api_key="your-api-key"))

# Basic text generation:
response = client.models.generate_content(
    model="gemini-2.5-flash",
    contents="Why is the sky blue?",
)
print(response.text)

# Streaming:
for chunk in client.models.generate_content_stream(
    model="gemini-2.5-flash",
    contents="Tell me a story",
):
    print(chunk.text, end="")

# Tool/Function calling:
schedule_meeting_function = {
    "name": "schedule_meeting",
    "description": "Schedules a meeting with specified attendees.",
    "parameters": {
        "type": "object",
        "properties": {
            "attendees": {"type": "array", "items": {"type": "string"}},
            "date": {"type": "string"},
            "time": {"type": "string"},
            "topic": {"type": "string"},
        },
        "required": ["attendees", "date", "time", "topic"],
    },
}

tools = types.Tool(function_declarations=[schedule_meeting_function])
config = types.GenerateContentConfig(tools=[tools])

response = client.models.generate_content(
    model="gemini-2.5-flash",
    contents="Schedule a meeting with Bob and Alice tomorrow at 2 PM.",
    config=config,
)

# Image generation:
response = client.models.generate_content(
    model="gemini-2.5-flash-image",
    contents=["Create a picture of a futuristic city"],
)

# Save generated image
from io import BytesIO
from PIL import Image

for part in response.candidates[0].content.parts:
    if part.inline_data is not None:
        image = Image.open(BytesIO(part.inline_data.data))
        image.save("generated_image.png")

Added in langsmith 0.4.33

Initial beta release of Google Gemini wrapper.

wrap_openai

wrap_openai(
    client: C,
    *,
    tracing_extra: TracingExtra | None = None,
    chat_name: str = "ChatOpenAI",
    completions_name: str = "OpenAI",
) -> C

Patch the OpenAI client to make it traceable.

Supports

• Chat and Responses API's
• Sync and async OpenAI clients
• create and parse methods
• With and without streaming

PARAMETER	DESCRIPTION
client	The client to patch. TYPE: C
tracing_extra	Extra tracing information. TYPE: TracingExtra | None DEFAULT: None
chat_name	The run name for the chat completions endpoint. TYPE: str DEFAULT: 'ChatOpenAI'
completions_name	The run name for the completions endpoint. TYPE: str DEFAULT: 'OpenAI'

RETURNS	DESCRIPTION
C	The patched client.

Example

import openai
from langsmith import wrappers

# Use OpenAI client same as you normally would.
client = wrappers.wrap_openai(openai.OpenAI())

# Chat API:
messages = [
    {"role": "system", "content": "You are a helpful assistant."},
    {
        "role": "user",
        "content": "What physics breakthroughs do you predict will happen by 2300?",
    },
]
completion = client.chat.completions.create(
    model="gpt-4o-mini", messages=messages
)
print(completion.choices[0].message.content)

# Responses API:
response = client.responses.create(
    model="gpt-4o-mini",
    messages=messages,
)
print(response.output_text)

Behavior changed in langsmith 0.3.16

Support for Responses API added.