Skip to content

LocoLLM Documentation

Frontier AI on a budget. Crazy, right?

LocoLLM is an open-source experiment in building routed, task-specific language models that run entirely on consumer hardware. One base model, many specialist adapters, zero cloud dependencies.


Understand the project’s goals, hardware, and philosophy.

Why LocoLLM · FAQ · Meet the Lab

Hands-on guides for training adapters, benchmarking, and evaluation.

Adapter Guide · Adapter Primer

Architecture decisions, model selection rationale, and research direction.

Architecture · Research Roadmap · Architecture Vision

Interactive results: quality, speed, and efficiency across 14 models x 8 quant levels.

Overview

Project ideas for capstone students and contributors.

Project Ideas

Recorded technical decisions and the reasoning behind them.

ADR Overview


Landing Page · GitHub · MIT License