Nocturne

Inference on idle Apple Silicon Macs

Nocturne is a self-hosted, OpenAI-compatible inference project using idle Apple Silicon Macs. Each request runs on one worker, not a distributed cluster of Macs.

Choose a Nocturne instance and obtain its public API base URL and API key from its operator. Check its available models and status before sending work: workers and availability can change.

Install the OpenAI Python client in your own environment. Set NOCTURNE_BASE_URL to the chosen instance's API base URL and NOCTURNE_API_KEY to your own key. Select a model returned by that instance, not a hardcoded model name.

The upstream project describes Lightning payment support. Availability, credit and payment arrangements are instance-specific; review the chosen operator's terms before use. Hbit does not collect keys, submit prompts, buy credit or initiate payments.

This is a client introduction, not a hosted-service availability promise. Consult the current upstream README for deployment, worker setup and billing details.

Nocturne source and current documentation

Apple Silicon worker setup

List the chosen instance's models (run on your own machine)

# Install in your own Python environment: python -m pip install openai
# Set NOCTURNE_BASE_URL and NOCTURNE_API_KEY privately first.
import os
from openai import OpenAI

client = OpenAI(
    base_url=os.environ["NOCTURNE_BASE_URL"],
    api_key=os.environ["NOCTURNE_API_KEY"],
)
for model in client.models.list().data:
    print(model.id)