SillyTavern

A local LLM frontend that connects to local backends and cloud APIs, with lorebooks, image generation and text-to-speech.

Alternative to:

SillyTavern is a locally installed interface for AI hobbyists who want control over prompts and access to text, image and voice models in one place. It's an independently developed fork of TavernAI and connects to a separate model backend or API provider.

Model connections: Use local backends such as KoboldAI/CPP, Ooba and Tabby, or cloud providers including OpenAI, OpenRouter, Claude, Mistral and NovelAI.
Prompts and lorebooks: Shape model responses with detailed prompt controls and supply background information through WorldInfo lorebooks.
Images and voice: Generate images through Automatic1111 or ComfyUI integrations and use text-to-speech models for spoken output.
Interface customization: Choose Visual Novel Mode, use automatic translation, customize the mobile-friendly interface and add capabilities through third-party extensions.

The frontend runs on Windows, macOS, Linux and Android through Termux, with Docker support. It's free and open source under AGPL-3.0. The project offers no hosted service and doesn't programmatically track user data. The interface runs locally; cloud API requests go to the provider you choose.

The frontend itself has modest hardware needs. Running models locally has separate requirements that depend on the model and backend. The project's recommendation for local inference is an NVIDIA 3000-series GPU with at least 6 GB of VRAM.