LoLLMS WebUI is a Python-based web UI for multi-model LLM and multimodal tasks, with manual/automatic installation options, multiple bindings, and a Smart Routing feature for model selection based on cost and speed.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
LoLLMS WebUI is a web user interface for local, single-user/multi-model and multimodal AI systems. It supports binding to multiple local and remote models and provides a UI with light/dark modes, personality customization, and integration with various services.
How it works
The project exposes a web interface to manage and run multiple LLM bindings and modalities. It includes a core library, submodules for bindings, and a smart routing mechanism to optimize model selection by cost and speed. It supports Docker, conda, and manual virtual environments for setup.
Getting started
Automatic installation references the scripts folder with lollms_installer.bat (Windows), lollms_installer.sh (Linux), and lollms_installer_macos.sh (Mac).
Manual install steps include:
- Ensure Python 3.11 is installed
- Clone the repository with submodules
git clone --recursive https://github.com/ParisNeo/lollms-webui.git
cd lollms-webui
git submodule update --init --recursive
- Create and activate a virtual environment
python -m venv venv
- Install requirements
pip install --upgrade pip
pip install -r requirements.txt
pip install -e lollms_core
- Create global_paths_cfg.yaml
cat > global_paths_cfg.yaml << EOL
lollms_path: $(pwd)/lollms_core/lollms
lollms_personal_path: $(pwd)/personal_data
EOL
- Install desired bindings using provided commands (example for ollama):
python zoos/bindings_zoo/ollama_ai/__init__.py
- Run the server
python app.py
Recent releases
Latest releases include:
- v14 V14 (Saïph) (2024-11-11): Enhanced UI with themes system, localization of personalities, bug fixes in UI, and security fixes.
- v13 V13 (feather) (2024-10-07): Light as a feather; removed conda dependence.
- v12 V12 (Strawberry) (2024-09-01): New rag system, improved smart routing for LLM selection based on prompt complexity, UI upgrades, and Apps zoo for building/testing/sharing apps.
- v9.8 (2024-05-20): Fixed security issues and enhanced UI accessibility for mounted personalities.
- v9.6 (2024-04-21): Enhanced UI, generation process insights, security, and code compression tools; personalities can load models.
Full changelog is available at the repository page.
Traction
Stars: 4786; Forks: 585; Open issues: 184.
Behind the repo
The repo is maintained by ParisNeo; no linked startup/company information is provided beyond the author and repository metadata.
Caveats
License: Apache-2.0. The project notes that remote access security considerations apply; there is a security warning about lack of built-in user authentication and guidance for securing remote access. Docker is supported; Docker run example provided:
docker build -t lollms-webui-app .
docker run -d -p 9600:9600 --name lollms-webui-container lollms-webui-app