Provides Python types, Pydantic models, and helpers to extend the OpenAI client for the Braintrust proxy, improving developer experience, typechecking, and editor integration.
Project description
Braintrust AI Proxy Python Client
The braintrust-proxy package provides Python types, Pydantic models, and helper functions to simplify interaction with the Braintrust AI proxy. By providing rich Python types that extend the OpenAI client types, it significantly enhances compatibility and developer experience, enabling features like typechecking and improved editor autocomplete when working with proxy-specific extensions (such as reasoning).
Installation
You can install the package using uv (recommended) or pip:
Using uv:
uv pip install braintrust-proxy
Using pip:
pip install braintrust-proxy
Usage
The package provides enhanced types and helper functions, such as as_openai_chat_message_param, to work with messages that might include proxy-specific extensions (like reasoning) while maintaining compatibility with standard OpenAI client inputs.
import os
from openai import OpenAI
import json
from braintrust_proxy import as_openai_chat_message_param
# Initialize the OpenAI client pointing to the Braintrust proxy
# Ensure BRAINTRUST_API_URL and BRAINTRUST_API_KEY environment variables are set
client = OpenAI(
# if self hosting, otherwise use the hosted API
base_url=f"{os.getenv('BRAINTRUST_API_URL') or 'https://api.braintrust.dev'}/v1/proxy",
api_key=os.getenv("BRAINTRUST_API_KEY"),
)
# Example using the helper function with a message that includes 'reasoning'
response = client.chat.completions.create(
model="claude-3-7-sonnet-latest",
reasoning_effort=None,
stream=False,
messages=[
{
"role": "user",
"content": "How many rs in 'ferrocarril'",
},
as_openai_chat_message_param(
{
"role": "assistant",
"content": "There are 4 letter 'r's in the word \"ferrocarril\".",
"reasoning": [
{
"id": "",
"content": "To count the number of 'r's in the word 'ferrocarril', I'll just go through the word letter by letter.\n\n'ferrocarril' has the following letters:\nf-e-r-r-o-c-a-r-r-i-l\n\nLooking at each letter:\n- 'f': not an 'r'\n- 'e': not an 'r'\n- 'r': This is an 'r', so that's 1.\n- 'r': This is an 'r', so that's 2.\n- 'o': not an 'r'\n- 'c': not an 'r'\n- 'a': not an 'r'\n- 'r': This is an 'r', so that's 3.\n- 'r': This is an 'r', so that's 4.\n- 'i': not an 'r'\n- 'l': not an 'r'\n\nSo there are 4 'r's in the word 'ferrocarril'.",
},
],
}
),
{
"role": "user",
"content": "How many e in what you said?",
},
],
)
print(
json.dumps(
{
"message": response.choices[0].message.dict(),
"reasoning": getattr(response.choices[0].message, "reasoning", None),
},
indent=2,
)
)
Streaming
import os
from openai import OpenAI
import json
from braintrust_proxy import as_openai_chat_message_param, from_openai_chat_completion_choice_delta
# Initialize the OpenAI client pointing to the Braintrust proxy
# Ensure BRAINTRUST_API_URL and BRAINTRUST_API_KEY environment variables are set
client = OpenAI(
# if self hosting, otherwise use the hosted API
base_url=f"{os.getenv('BRAINTRUST_API_URL') or 'https://api.braintrust.dev'}/v1/proxy",
api_key=os.getenv("BRAINTRUST_API_KEY"),
)
stream = client.chat.completions.create(
model="claude-3-7-sonnet-latest",
reasoning_effort="high",
stream=True,
messages=[
{
"role": "user",
"content": "How many rs in 'ferrocarril'",
},
as_openai_chat_message_param(
{
"role": "assistant",
"content": "There are 4 letter 'r's in the word \"ferrocarril\".",
"reasoning": [
{
"id": "",
"content": "To count the number of 'r's in the word 'ferrocarril', I'll just go through the word letter by letter.\n\n'ferrocarril' has the following letters:\nf-e-r-r-o-c-a-r-r-i-l\n\nLooking at each letter:\n- 'f': not an 'r'\n- 'e': not an 'r'\n- 'r': This is an 'r', so that's 1.\n- 'r': This is an 'r', so that's 2.\n- 'o': not an 'r'\n- 'c': not an 'r'\n- 'a': not an 'r'\n- 'r': This is an 'r', so that's 3.\n- 'r': This is an 'r', so that's 4.\n- 'i': not an 'r'\n- 'l': not an 'r'\n\nSo there are 4 'r's in the word 'ferrocarril'.",
},
],
}
),
{
"role": "user",
"content": "How many e in what you said?",
},
],
)
for event in stream:
delta = from_openai_chat_completion_choice_delta(event.choices[0].delta)
print(
json.dumps(
{"content": delta.content, "reasoning": delta.reasoning.dict() if delta.reasoning else None}, indent=2
)
)
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file braintrust_proxy-0.1.0.tar.gz.
File metadata
- Download URL: braintrust_proxy-0.1.0.tar.gz
- Upload date:
- Size: 3.0 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.6.10
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
5e37a779cefacd2871be2578bfbbbb10024b38c015b086f3c4520066a96b16ab
|
|
| MD5 |
0f20d97b65ad2025c54248ffdfcf9ecf
|
|
| BLAKE2b-256 |
7cca7c7193bdc7db69dbbfa7354cdfe3be6605b822dd29110484cd1b83009075
|
File details
Details for the file braintrust_proxy-0.1.0-py3-none-any.whl.
File metadata
- Download URL: braintrust_proxy-0.1.0-py3-none-any.whl
- Upload date:
- Size: 3.9 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via: uv/0.6.10
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
198e0f8eb2078f28a8a1724e0257ec094079bfeb252960a13007a2c51484fc33
|
|
| MD5 |
3ba029a601dc3c4b7608c92e6065e1ec
|
|
| BLAKE2b-256 |
272e6a5a213922aca96816a1edbca88d6ac9cfce0986f03d2b87a74ec8a2958d
|