How to trace Azure OpenAI with HoneyHive

Azure OpenAI provides OpenAI models hosted on Azure infrastructure. HoneyHive integrates with Azure OpenAI using the same OpenInference OpenAI instrumentor, automatically capturing all API calls, tool use, and token usage.

Quick Start

Add HoneyHive tracing in just 4 lines of code. Azure OpenAI uses the same openai Python package and the same instrumentor as standard OpenAI.

To see where to initialize the tracer for your environment, including AWS Lambda and long-running servers, see Tracer Initialization.

pip install "honeyhive[openinference-azure-openai]"

# Or install separately
pip install honeyhive openinference-instrumentation-openai openai azure-identity

import os
from honeyhive import HoneyHiveTracer
from openinference.instrumentation.openai import OpenAIInstrumentor

tracer = HoneyHiveTracer.init(api_key=os.getenv("HH_API_KEY"))
OpenAIInstrumentor().instrument(tracer_provider=tracer.provider)

# Your existing Azure OpenAI code works unchanged

Azure OpenAI uses the standard openai Python package with the AzureOpenAI client class. The same OpenAIInstrumentor works for both.

Tested Versions

HoneyHive’s Azure OpenAI integration is tested against the following versions on PyPI, as of May 2026. Newer patch releases are generally safe; if you hit an issue, pin to these versions to reproduce a known-good configuration.

Package	Version
`openinference-instrumentation-openai`	`>= 0.1.0`
`openai`	`>= 1.0.0`
`azure-identity`	`>= 1.12.0`

Requires Python 3.11+.

What Gets Traced

The instrumentor automatically captures:

Chat completions - client.chat.completions.create() with inputs, outputs, and token usage
Tool / function calls - Each tool call with arguments and results
Embeddings - client.embeddings.create() requests
Streaming responses - Streamed completions with aggregated tokens

No manual instrumentation required.

Example: Basic Chat Completion

import os
from honeyhive import HoneyHiveTracer
from openinference.instrumentation.openai import OpenAIInstrumentor
from openai import AzureOpenAI

tracer = HoneyHiveTracer.init(api_key=os.getenv("HH_API_KEY"))
OpenAIInstrumentor().instrument(tracer_provider=tracer.provider)

client = AzureOpenAI(
    api_key=os.getenv("AZURE_OPENAI_API_KEY"),
    api_version="2024-10-21",
    azure_endpoint=os.getenv("AZURE_OPENAI_ENDPOINT", ""),
)

response = client.chat.completions.create(
    model="gpt-4o-mini",  # Your deployment name
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is the capital of France?"},
    ],
)
print(response.choices[0].message.content)

Environment Configuration

# HoneyHive configuration
export HH_API_KEY="your-honeyhive-api-key"

# Azure OpenAI configuration
export AZURE_OPENAI_API_KEY="your-azure-openai-key"
export AZURE_OPENAI_ENDPOINT="https://your-resource.openai.azure.com/"

Troubleshooting

Traces not appearing

Check environment variables - Ensure HH_API_KEY is set
Pass the tracer provider - The instrumentor must receive tracer_provider=tracer.provider:

from honeyhive import HoneyHiveTracer
from openinference.instrumentation.openai import OpenAIInstrumentor

tracer = HoneyHiveTracer.init(api_key=os.getenv("HH_API_KEY"))

# ✅ Correct - pass tracer_provider
OpenAIInstrumentor().instrument(tracer_provider=tracer.provider)

# ❌ Wrong - missing tracer_provider
OpenAIInstrumentor().instrument()

Initialize before making calls - Call instrument() before creating the AzureOpenAI client

Deployment name vs model name

In Azure OpenAI, the model parameter is your deployment name, not the underlying model name:

response = client.chat.completions.create(
    model="my-gpt4-deployment",  # This is your Azure deployment name
    messages=[{"role": "user", "content": "Hello"}],
)

OpenAI Integration

Standard OpenAI API integration

Enrich Your Traces

Add user IDs and custom metadata to traces

Distributed Tracing

Trace calls across service boundaries

Using Traceloop (OpenLLMetry) Instead

If your project already uses Traceloop / OpenLLMetry, you can use its OpenAI instrumentor instead of OpenInference. The setup is identical - only the install and import paths differ.

pip install "honeyhive[traceloop-azure-openai]"

# Or install separately
pip install honeyhive opentelemetry-instrumentation-openai openai azure-identity

import os
from honeyhive import HoneyHiveTracer
from opentelemetry.instrumentation.openai import OpenAIInstrumentor
from openai import AzureOpenAI

tracer = HoneyHiveTracer.init(api_key=os.getenv("HH_API_KEY"))
OpenAIInstrumentor().instrument(tracer_provider=tracer.provider)

client = AzureOpenAI(
    api_key=os.getenv("AZURE_OPENAI_API_KEY"),
    api_version="2024-10-21",
    azure_endpoint=os.getenv("AZURE_OPENAI_ENDPOINT", ""),
)

response = client.chat.completions.create(
    model="gpt-4o-mini",  # Your deployment name
    messages=[{"role": "user", "content": "What is the capital of France?"}],
)
print(response.choices[0].message.content)

Tested Versions

Package	Version
`opentelemetry-instrumentation-openai`	`>= 0.58.0, < 1.0.0`
`openai`	`>= 1.0.0`
`azure-identity`	`>= 1.12.0`

Agent Frameworks & Platforms

Model Providers

AI Gateways

Coding Agents

Evaluation

How to trace Azure OpenAI with HoneyHive

Quick Start

Tested Versions

What Gets Traced

Example: Basic Chat Completion

Environment Configuration

Troubleshooting

Traces not appearing

Deployment name vs model name

OpenAI Integration

Enrich Your Traces

Distributed Tracing

Using Traceloop (OpenLLMetry) Instead

Tested Versions

Resources

​Quick Start

​Tested Versions

​What Gets Traced

​Example: Basic Chat Completion

​Environment Configuration

​Troubleshooting

​Traces not appearing

​Deployment name vs model name

​Related

OpenAI Integration

Enrich Your Traces

Distributed Tracing

​Using Traceloop (OpenLLMetry) Instead

​Tested Versions

​Resources

Quick Start

Tested Versions

What Gets Traced

Example: Basic Chat Completion

Environment Configuration

Troubleshooting

Traces not appearing

Deployment name vs model name

Related

Using Traceloop (OpenLLMetry) Instead

Tested Versions

Resources