requirements.txt

flask==3.0.3
flask-cors==6.0.0
gunicorn==22.0.0
gevent==24.2.1

# llama.cpp python bindings (use prebuilt CPU wheels)
--extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu
llama-cpp-python==0.2.90

# for downloading the GGUF from the Hub at startup
huggingface_hub>=0.23.0

# other stuff
torch
transformers
pydantic
python-dotenv
fastapi
uvicorn[standard]