zyx

---

An Easier Way to Work With LLMs

zyx is a python package built with the sole purpose of providing an intutive and simple API when working and building with LLMs. It aims to remove the current boilerplate of most LLM libraries, through quick-use functions and modules.

The point of zyx is not to be a flashy new library, everything here can be done without much effort, it just helps you do them quicker. zyx is a developer focused library, it's built so your ideas come out easier.

The framework is built off of incredibly well-built libraries, most notably:

• Instructor
• LiteLLM
• Chroma
• Pydantic

---

Quick Start

Installation

Install zyx using pip:

pip install --upgrade zyx

---

import zyx

def my_favorite_food(name : str) -> str:
    """Returns the user's favorite food."""
    return "pizza"

zyx.completion(
    "What is my favorite food?",
    model = "gpt-4o-mini", # Any LiteLLM model is compatible
    tools = [my_favorite_food], # Any function, basemodel or openai tool
    run_tools = True # Optional: automatically executes tools
    # response_model = BaseModel # Optional: Pydantic model for structured outputs
    # mode = "tool_call" # Optional: Instructor completion mode (refer to instructor docs)
)

# "Your favorite food is pizza!"