Anthropic Tracing

Instrument LLM calls made using Anthropic's SDK and view traces in Arize.

Anthropic is a leading provider for state-of-the-art LLMs. The Anthropic SDK can be instrumented using the openinference-instrumentation-anthropic package. Traces can be viewed in Arize.

Launch Arize

To get started, sign up for a free Arize account and get your Space ID and API Key.

Install

pip install openinference-instrumentation-anthropic anthropic arize-otel opentelemetry-sdk

API Key Setup

Ensure your Anthropic API key is configured. The SDK typically uses the ANTHROPIC_API_KEY environment variable:

export ANTHROPIC_API_KEY='your-anthropic-api-key'

Setup

Use the register function to connect your application to Arize and instrument the Anthropic client:

# Import open-telemetry dependencies
from arize.otel import register

# Setup OTel via our convenience function
tracer_provider = register(
    space_id = "your-space-id", # in app space settings page
    api_key = "your-api-key", # in app space settings page
    project_name = "your-project-name", # name this to whatever you would like
)

# Import openinference instrumentor to map Anthropic traces to a standard format
from openinference.instrumentation.anthropic import AnthropicInstrumentor

# Turn on the instrumentor
AnthropicInstrumentor().instrument(tracer_provider=tracer_provider)

Run Anthropic

A simple Anthropic application that is now instrumented:

import anthropic

client = anthropic.Anthropic() # The client will use the ANTHROPIC_API_KEY environment variable

message = client.messages.create(
    model="claude-3-5-sonnet-20240620",
    max_tokens=1000,
    temperature=0,
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "Why is the ocean salty?"
                }
            ]
        }
    ]
)
print(message.content)

Observe

Now that you have tracing setup, all invocations will be streamed to your Arize account for observability and evaluation.

Resources:

• Example Messages

• Example Tool Calling

• OpenInference package