Guess Llama - A game for local Vision LLM

I've been working on a project I call Guess Llama. The concept is based on the old ' Guess Who? ' game. 'Guess Llama' uses a vision LLM backend such as llama.cpp 's llama-server to generate and play the game. It currently uses stable-diffusion.cpp 's sd-server or Openrouter.ai image generating models to generate the images. You can enter any 'theme' for the game, or ask the bot to generate one. Such as 'cat', 'llama', 'capybara', 'clown', 'space alien', etc. The bot suggests 8 items that can go with the theme.