LumiChats Offline LLM

Overview

LumiChats Offline is built on top of GPT4All by Nomic AI — the leading open-source local LLM runner — with LumiChats branding, enhanced privacy defaults, a refined dark theme, and our custom fine-tuned models optimised for real-world tasks.

Download & Install

System Requirements

Quickstart

1. 1

   Download the app

   Click the download button above and extract the zip folder anywhere on your PC.

2. 2

   Run chat.exe

   Double-click chat.exe inside the extracted folder. No installation wizard needed.

3. 3

   Install a model

   Click Models in the sidebar → + Add Model → choose and download a model (~1–5 GB).

4. 4

   Start chatting

   Click Chats → + New Chat → select your model → type your message and press Enter.

Installing Models

LumiChats Offline supports any GGUF-format model. Models are downloaded to your device and run entirely locally. Browse and install models directly from inside the app.

LocalDocs (RAG)

LocalDocs lets you chat with your own files — PDFs, Word documents, text files — using local retrieval-augmented generation (RAG). Your files are embedded locally using the bundled nomic-embed-text model. Nothing is sent to any server.

1. 1

   Click LocalDocs in the sidebar

2. 2

   Click + Add Collection and name your collection

3. 3

   Select a folder on your PC that contains your documents

4. 4

   Click Create Collection — embedding runs locally in the background

5. 5

   When the green Ready indicator appears, your docs are indexed

6. 6

   In any chat, open LocalDocs from the top-right button to activate it

7. 7

   Ask questions — the model will reference your documents for answers

Settings Reference

Build from Source

For developers who want to build LumiChats Offline from the source code. Requires Windows with Visual Studio 2022.

Prerequisites

• Visual Studio 2022 with Desktop C++ workload
• CMake 3.21+
• Qt 6.8+ (installed via aqtinstall)
• Python 3.10+

Clone & Build

git clone https://github.com/adityajhakumar/LumiChats-Offline-LLM.git
cd LumiChats-Offline-LLM

# Install Qt
pip install aqtinstall
aqt install-qt windows desktop 6.8.3 win64_msvc2022_64 -O C:\Qt

# Configure (run in x64 Native Tools Command Prompt)
cmake -S gpt4all-chat -B build ^
  -DCMAKE_BUILD_TYPE=Release ^
  -DCMAKE_PREFIX_PATH="C:\Qt\6.8.3\msvc2022_64" ^
  -DLLMODEL_CUDA=OFF -DLLMODEL_VULKAN=OFF

cmake --build build --config Release --parallel

# Deploy Qt dependencies
C:\Qt\6.8.3\msvc2022_64\bin\windeployqt.exe build\bin\Release\chat.exe

LumiChats Models

We have fine-tuned a suite of open-source models specifically for LumiChats use cases. All models are free to use on Hugging Face and can be downloaded into LumiChats Offline via the Models tab.

FAQ

Attribution & License

LumiChats Offline is a derivative work of GPT4All by Nomic AI and contributors, used under the MIT License. We are deeply grateful to the Nomic AI team for building and open-sourcing this foundational technology.

@misc{gpt4all,
  author    = {Anand, Yuvanesh and Nussbaum, Zach and Duderstadt, Brandon
               and Schmidt, Benno and others},
  title     = {GPT4All: An Ecosystem of Open Source Compressed LLMs},
  year      = {2023},
  publisher = {Nomic AI},
  url       = {https://github.com/nomic-ai/gpt4all}
}

LumiChats-specific modifications © 2026 LumiChats · Original GPT4All code retains its copyright by Nomic AI and contributors. Model weights downloaded through the app are subject to their respective licenses.