Yollama：主打极致隐私与离线运行的 Ollama 分支

Yollama Development | May-2025 to September-2026

Enjoy true data privacy with real offline Freedom for your LLM AI models

Welcome!

Thank you for taking the time to read this documentation of Yollama!
Yollama is a fork of Ollama that is geared directly for a "no frills" and "extreme privacy + security" perspective as its main focus.

Currently, Ollama has many back-door connections, hidden remote servers, and countless other aspects of the system that are far from "local only" model running. This project, Yollama, (stands for Your-Ollama) aims to remedy what Ollama quietly forces onto their users.

Far from secure- Ollama enacts many hidden and not-made-public functionalities of their program that hinders a user's ability to guide and run an LLM AI model to the point or aspect that they desire. Yollama project in contrast aims at taking back control for the user, ensuring there are no Automated-Updates, Remote-connections, Remotely defined/used "Model-Suggestions", and no other underlying operations that could possibly cause a supposed "local-only" runtime from affecting the user of this project.

Yollama focuses on clean and seamless operation including removing any hindering aspects and adding newly rewritten handling (Yollama does not include a UI, MLX-Runner, Proprietary Templates, or "APP" as the LLM-server and model-runner are the focus). With all of the hidden network outreach and remote overrides removed from the source- users will no longer see AI inference quality wildly variate as they interact with their LLM AI model(s), nor will a user see the AI responses simply fail when disconnecting from their network (seen currently in Ollama when a model is loaded and then the host machine is disconnected from the network during any portion of the LLM model attempting to respond as the response will fail). Yollama project also boasts faster compile times, such as a blazing ~30 seconds using most any modern CPU (e.g. Core-i7 10th gen and higher). As an example: an Intel Xeon Silver 4216 compiled this project within 1-minute.

Get started

TODO: Please wait for updates, they are on their way.

REST API

Yollama has a REST API for running and managing models.

Supported backends

• llama.cpp project founded by Georgi Gerganov.