> ## Documentation Index
> Fetch the complete documentation index at: https://docs.fallom.com/llms.txt
> Use this file to discover all available pages before exploring further.

# OpenAI (Python)

> Trace OpenAI API calls in Python

## Installation

```bash theme={null}
pip install fallom openai
```

## Quick Start

<Warning>
  **Import order matters!** Initialize Fallom **before** importing OpenAI.
</Warning>

```python theme={null}
import os

# Import and initialize Fallom FIRST
import fallom
fallom.init(api_key=os.environ["FALLOM_API_KEY"])

# NOW import OpenAI
from openai import OpenAI
client = OpenAI()

# Set session context (config_key, session_id, customer_id)
fallom.trace.set_session("my-app", "session-123", customer_id="user-456")

# Use as normal - automatically traced!
response = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Hello!"}]
)

print(response.choices[0].message.content)
```

## Why Import Order Matters

Fallom uses OpenTelemetry auto-instrumentation which patches the OpenAI library when initialized. If you import OpenAI first, the instrumentation can't patch it properly.

```python theme={null}
# ❌ WRONG - OpenAI imported before Fallom init
from openai import OpenAI
import fallom
fallom.init()  # Too late - OpenAI already loaded

# ✅ CORRECT - Fallom init before OpenAI import
import fallom
fallom.init()
from openai import OpenAI  # Now it gets patched
```

## Streaming

```python theme={null}
import fallom
fallom.init(api_key="your-fallom-api-key")

from openai import OpenAI
client = OpenAI()

fallom.trace.set_session("my-agent", "session-123")

stream = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "Write a poem."}],
    stream=True
)

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

## With Azure OpenAI

```python theme={null}
import fallom
fallom.init(api_key="your-fallom-api-key")

from openai import AzureOpenAI

client = AzureOpenAI(
    api_key=os.environ["AZURE_OPENAI_API_KEY"],
    api_version="2024-02-01",
    azure_endpoint=os.environ["AZURE_OPENAI_ENDPOINT"]
)

fallom.trace.set_session("my-agent", "session-123")

response = client.chat.completions.create(
    model="gpt-4o",  # Your deployment name
    messages=[{"role": "user", "content": "Hello!"}]
)
```

## Model A/B Testing

```python theme={null}
import fallom
from fallom import models

fallom.init(api_key="your-fallom-api-key")

from openai import OpenAI
client = OpenAI()

# Get assigned model for this session
model_id = models.get("my-experiment", "session-123", fallback="gpt-4o-mini")

fallom.trace.set_session("my-experiment", "session-123")

response = client.chat.completions.create(
    model=model_id,  # Uses A/B test assigned model
    messages=[{"role": "user", "content": "Hello!"}]
)
```

## What Gets Traced

| Field       | Description                       |
| ----------- | --------------------------------- |
| Model       | `gpt-4o`, `gpt-4o-mini`, etc.     |
| Duration    | Total request time (ms)           |
| Tokens      | Prompt, completion, cached tokens |
| Cost        | Calculated from token usage       |
| Prompts     | All input messages                |
| Completions | Model response                    |
| Session     | Your config key + session ID      |

## Next Steps

<CardGroup cols={2}>
  <Card title="Model A/B Testing" icon="flask" href="/model-testing">
    Test different models in production.
  </Card>

  <Card title="Prompt Management" icon="pen-fancy" href="/prompt-testing">
    Manage and version your prompts.
  </Card>
</CardGroup>