SillyTavern
sillytavern.app
- Free plan
- Open source
- Windows
- macOS
- Linux
- Web
- Android
- Browser extension
- API
- Self-hosted
What SillyTavern is
SillyTavern is a locally installed interface for interacting with text-generation LLMs, image-generation engines, and text-to-speech voice models. Users can create, import, and switch between character cards, then resume or review chat histories. The interface includes group chats, lorebooks, user personas, prompt controls, chat commands, scripting, and built-in retrieval-augmented generation. Extensions can add character sprites, summaries and translation, image generation, text-to-speech, and web search. SillyTavern supports local and cloud model backends, including KoboldCpp, llama.cpp, Ollama, OpenAI, Claude, Mistral, OpenRouter, and AI Horde. It is free and open source under the AGPL-3.0 license, with no online or hosted service. The project says it does not programmatically track user data, though cloud model providers may log information sent to them. Local use requires a system that supports NodeJS 20 or higher; the documentation recommends an NVIDIA 3000-series GPU with at least 6GB of VRAM for local inference. The project says performance can degrade with character libraries above 1,000 characters.
Who it suits
SillyTavern is aimed at AI hobbyists and power users who want control over their LLM prompts. It suits users comfortable with local installation and, for local inference, the documented system requirements.
Good to know, and what to watch
- Free and open source under AGPL-3.0
- Supports local and cloud model backends
- Character chats include resumable histories
- Extensions add image, voice, and web-search capabilities
- Requires a system supporting NodeJS 20 or higher
- Local inference recommendations include a 6GB VRAM GPU
- Performance can degrade above 1,000 characters
- Cloud model providers may log sent data
Plans
Every plan and priceSillyTavern
Free plan
The verdict
SillyTavern gives users a locally installed interface with extensive prompt and character-chat controls, plus connections to local and cloud models. It is a better fit for users prepared to manage a local setup than for anyone seeking a hosted service.
What the maker publishes
- API integrations
- Documented connectors include local KoboldCpp, llama.cpp, Ollama, Oobabooga and TabbyAPI plus cloud services such as OpenAI, Claude, Gemini, Mistral, OpenRouter, DeepSeek, Cohere, Perplexity and others. source
- API key storage
- SillyTavern stores API keys in a secrets.json file in the user data directory and hides them in the interface by default after saving and refreshing. source
- API trade-offs
- Local APIs are free and private but can require complex setup and 5–50 GB model downloads, while cloud APIs are generally stronger but usually paid and may filter content. source
- Audience
- The project describes itself as a tool for savvy AI hobbyists and power users who want control over their LLM prompts. source
- Backend dependency
- SillyTavern supplies the interface and requires an LLM backend such as AI Horde, OpenAI-compatible services, KoboldAI or Tabby. source
- Character cards
- Character cards store prompts that define an AI character's behavior and support persistent conversations. source
- Character chats
- Users can create, import, and switch between character cards, with chat histories they can resume or review. source
- Chat features
- Features include group chats, lorebooks, user personas, prompt controls, chat commands, scripting, and built-in retrieval-augmented generation. source
- Commands and scripting
- The product includes a chat-command subsystem and its own STscript scripting language. source
- Developer extensibility
- STscript, function calling, browser UI extensions and Node.js server plugins allow scripts, event hooks, new endpoints and external-provider integrations. source
- Extensions
- Built-in or downloadable extensions provide sprites, summaries, translation, Stable Diffusion/FLUX/DALL-E images, text-to-speech and web search. source
- Founded
- 2023 source
- Group chats
- Group-chat rooms allow multiple character bots to converse with the user or one another. source
- Image generation integrations
- The project lists Stable Diffusion, FLUX, and DALL-E image generation among its extension capabilities. source
- Installation platforms
- Official guides cover Windows, Linux, macOS, Android through Termux and Docker, with release and staging branches. source
- Integrations
- SillyTavern supports local and cloud LLM backends, including KoboldCpp, llama.cpp, Ollama, OpenAI, Claude, Mistral, OpenRouter, and AI Horde. source
- Licence
- SillyTavern is released under the AGPL-3.0 licence. source
- Limit: context management
- An independent benchmark notes that lore consumes context budget and complex multi-character prompting can cause ordering and consistency problems. source
- Limit: interface
- Users report confusing menus, inconsistent settings locations, weak mobile usability and difficult card management. source
- Limit: setup
- Independent users describe installation, API selection and first-time configuration as technically demanding and time-consuming. source
- Local inference hardware
- The documentation recommends an NVIDIA 3000-series GPU with at least 6 GB VRAM for local model inference. source
- Local model requirements
- SillyTavern runs on systems that support NodeJS 20 or higher, and the documentation recommends an NVIDIA 3000-series GPU with at least 6GB of VRAM for local inference. source
- Media and extensions
- Extensions support character sprites, chat summarization and translation, image generation, text-to-speech, and web search. source
- Mobile limitation
- iPhone and iPad cannot run the complete application locally; Android Termux operation is possible but officially unsupported. source
Checked against the maker's pages on 14 Oct 2026.