OpenAI SDK (Python)

The XeroML OpenAI Python integration wraps the official OpenAI client and automatically traces all API calls — chat completions, embeddings, image generation, and more.

Installation

pip install xeroml

The OpenAI integration is included in the core xeroml package.

Setup

1. Set environment variables:

   XEROML_SECRET_KEY="sk-xm-..."
   XEROML_PUBLIC_KEY="pk-xm-..."
   XEROML_BASE_URL="https://cloud.xeroml.com"
   OPENAI_API_KEY="sk-..."

2. Replace your OpenAI import:

   # Before
   import openai
   
   # After
   from xeroml.openai import openai

3. Use normally — all calls are automatically traced:

   from xeroml.openai import openai
   
   response = openai.chat.completions.create(
       model="gpt-4o",
       messages=[
           {"role": "system", "content": "You are a helpful assistant."},
           {"role": "user", "content": "Explain XeroML in one sentence."},
       ],
   )

What Gets Captured

For each API call, XeroML automatically records:

• Model name and parameters (temperature, max_tokens, etc.)
• Input messages (prompt)
• Completion response
• Token usage (prompt tokens, completion tokens, total)
• Estimated cost
• Latency (time to first token, total response time)
• Finish reason

Adding User and Session Context

Use propagate_attributes() to attach user and session context to OpenAI calls:

from xeroml.openai import openai
from xeroml import propagate_attributes

def handle_chat(message: str, user_id: str, session_id: str) -> str:
    with propagate_attributes(user_id=user_id, session_id=session_id):
        response = openai.chat.completions.create(
            model="gpt-4o",
            messages=[{"role": "user", "content": message}],
        )
    return response.choices[0].message.content

Async Support

The integration works identically with the async OpenAI client:

from xeroml.openai import AsyncOpenAI

async def generate(prompt: str) -> str:
    client = AsyncOpenAI()
    response = await client.chat.completions.create(
        model="gpt-4o",
        messages=[{"role": "user", "content": prompt}],
    )
    return response.choices[0].message.content

Streaming

Streaming completions are fully supported. The trace is completed when the stream closes:

from xeroml.openai import openai

stream = openai.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Tell me a story"}],
    stream=True,
)

for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="")