Silly Tavern in 2025! (Patreon)

Video - https://youtu.be/bHqJ6mo1P-4

SillyTavern is a user interface you can install on your computer (and Android phones) that allows you to interact with LLMs and other backend APIs.

Create your own tavern filled with characters of your choice. Give them all personalities, avatars and even individual knowledge via the data bank. Send and receive images too, now also powered by ComfyUI! Lots has changed since my original video from a year ago - https://youtu.be/_2-KwnJPwfE - with extra features and support for many more APIs :)

Works best with 6+GB VRAM, especially if you want to generate images using ComfyUI or other local generators.

Ollama is now also supported, so if you have that already then you've got the beating heart of the tavern ready to go!

Installing via the Launcher now appears to work (for the Tavern itself, at least), however for the best Speech-to-Text, I found the Extras were still needed.

Commands to install Silly Extras

For fast whisper, the Extras repo allows for GPU usage. Even with the default turbo large model, it's almost instant compared to over 14 seconds without GPU acceleration.

git clone https://github.com/SillyTavern/SillyTavern-Extras.git

conda create --name silly-extras python=3.11

conda activate silly-extras

cd SillyTavern-Extras

pip install -r requirements.txt

pip install -r requirements-faster-whisper.txt

Start the server using:

python server.py --use-faster-whisper --cuda --enable-modules=whisper-stt

Links

Links to Silly Tavern and the example Live2D + VRM files used:

 https://github.com/SillyTavern/SillyTavern

https://github.com/SillyTavern/SillyTavern-Launcher

https://github.com/SillyTavern/SillyTavern-Extras

https://docs.sillytavern.app/

https://ollama.com/

Live2d example: https://booth.pm/en/items/3072192

VRM example: https://booth.pm/en/items/3853613