Local_LLM · run language models without the cloud.
A desktop application for installing and running Large Language Models locally on your machine. No internet required after setup. No API key, no subscription, no data leaving your computer.
LLMs on your machine, managed from a single window.
Local_LLM wraps llama.cpp — the high-performance LLM inference engine — in a desktop GUI. You get a single interface for downloading models, configuring parameters, and running inference, without touching a command line.
The main use case is privacy-sensitive prompting: when the content of your queries should not leave your machine. Financial analysis, internal documents, personal notes — anything where cloud API logging is a concern.
What Local_LLM does
- Download and manage LLM model files from a GUI — no manual file handling
- Configure inference parameters (temperature, context length, seed) per session
- Run text generation and chat-style inference entirely on-device
- No API key, no subscription, no usage limits beyond your hardware
- No internet connection required after the model is downloaded
- Built on llama.cpp: supports quantized GGUF models, runs on CPU and GPU
Setup
Configuração
1. Clone & install
1. Clonar & instalar
Requires Python 3.9+ and Git. Clone the repository and install dependencies:
Requer Python 3.9+ e Git. Clone o repositório e instale as dependências:
git clone https://github.com/Andrade020/Local_LLMpip install -r requirements.txt
2. Download a model
2. Baixar um modelo
Local_LLM uses GGUF-format models (the quantized format used by llama.cpp). The app's model manager lets you download models directly. Recommended starting point: Mistral 7B Q4_K_M or any model from the Hugging Face GGUF collection.
O Local_LLM usa modelos no formato GGUF (o formato quantizado do llama.cpp). O gerenciador de modelos do app permite baixar modelos diretamente. Ponto de partida recomendado: Mistral 7B Q4_K_M ou qualquer modelo da coleção GGUF do Hugging Face.
3. Run
3. Rodar
Launch the app with python main.py. Select the downloaded model in the GUI, adjust parameters (temperature, context length), and start the conversation. Internet connection required only for the initial model download.
Inicie o app com python main.py. Selecione o modelo baixado na GUI, ajuste os parâmetros (temperatura, tamanho do contexto) e comece a conversa. Conexão com internet necessária apenas para o download inicial do modelo.
Stack
Core
Python
GUI
tkinter
Inference engine
llama.cpp