Jupyter notebooks with Meridian

Install the Meridian Python package and the JupyterLab AI extension:

pip install meridian-sdk jupyterlab-ai

Set your API key as an environment variable, then use Meridian in any notebook cell:

import os
from meridian import Meridian

client = Meridian(api_key=os.environ["MERIDIAN_API_KEY"])

response = client.chat.completions.create(
    model="meridian-pro",
    messages=[{"role": "user", "content": "Explain backpropagation"}],
)
print(response.choices[0].message.content)

Meridian exposes an OpenAI-compatible endpoint. Swap the base URL and keep your existing notebook workflows:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.meridian.ai/v1",
    api_key="sk-meridian-...",
)

completion = client.chat.completions.create(
    model="meridian-pro",
    messages=[{"role": "user", "content": "Summarize this dataset"}],
)

After installing jupyterlab-ai, configure the Meridian provider in your JupyterLab settings. You can then invoke Meridian inline with %%ai magic commands or use the chat sidebar without leaving your notebook.