Sider
AssistantsRead pages with Chat, handle browser tasks with Hand and customize your browsing experience with Code. Sider adds an AI workspace to the sites and documents you already use.
Chat beside the sourceHand browser tasks
SILLYTAVERN / GIVE CONVERSATIONS A CONSISTENT ROLE AND CONTEXT
SillyTavern
A free, locally installed frontend for structured conversations with language models. SillyTavern adds character cards, personas, World Info, group chats and document-related context, while leaving model inference to a separately configured local or cloud backend.
SillyTavern official documentation screenshot showing model connection settings and conversation controls; interface exampleExplore the image ↓
What you can do with it
What this tool is for
SillyTavern is a locally installed interface for interacting with language models. It gives users detailed control over how a conversation is framed, what background information is supplied and which model backend produces the answer. It is especially useful for interactive fiction, role-based exercises and specialized assistants that need consistent instructions.
The application is not a model and does not provide unlimited inference by itself. Users connect a supported local or cloud backend. This distinction matters for cost, privacy and hardware: running the interface can be lightweight, while running a capable model on the same machine can require substantially more resources.
01
Develop an interactive story
02
Practice a role-based conversation
03
Create a specialized assistant
04
Compare model behavior
A closer look at the product
Official product image. Click to inspect the details.
Product interface and examples
Key features, practical tasks
Capabilities in practice
- Character cards: A character card stores the prompt material that defines a role or scenario. It can represent a fictional character or a task-specific assistant. Keep the essential behavior concise so it remains understandable and leaves room for the actual conversation.
- User personas: Personas provide information about the user's role in an interaction. Use them to keep context consistent across conversations, while avoiding unnecessary personal details that would then be sent to the configured model provider.
- World Info entries: World Info inserts relevant background when configured triggers activate. This helps maintain a larger setting or reference collection without placing every detail in every prompt, but activation rules and context budgets need to be understood and tested.
- Group conversations: Multiple character roles can participate in a group-chat workflow. This is useful for exploring contrasting viewpoints or fictional dialogue, but agreement between generated roles is not independent verification of a factual claim.
- Backend selection: The documentation covers supported local and hosted APIs. Choose a backend based on the model and deployment requirements, and remember that API compatibility does not guarantee the same behavior, context capacity or quality.
- Document and extension workflows: The project describes document-reference support and extensions for tasks such as summaries, speech or images. These capabilities may require additional models or services, so introduce them only when they solve a concrete need.
- Prompt and interface control: Generation settings, formatting options, commands and interface customization support advanced workflows. Change a small number of settings at a time and retain a working configuration so a later issue can be traced to its cause.
Get started, step by step
- 01
Install the stable route
- 02
Connect one backend
- 03
Build a minimal conversation setup
- 04
Expand gradually
A practical first workflow
A practical first workflow
- Install the stable route. Follow the official guide for your platform and use the recommended release branch for ordinary use. Confirm the current Node.js requirement and keep the installation separate from unrelated projects.
- Connect one backend. Choose a supported local model service or an authorized cloud provider. Verify the connection with a harmless short prompt and check billing and data handling before adding private context.
- Build a minimal conversation setup. Create a short character or task card and only a few World Info entries. Test the triggers and context behavior with simple questions so failures are easy to understand.
- Expand gradually. Add personas, group roles or extensions after the basic workflow is useful. Keep backups of cards and settings, and inspect what information is being included when a conversation behaves unexpectedly.
Example prompt or task: Act as a fictional museum guide in this educational scenario. Use only the supplied exhibit facts, ask one question at a time, and clearly distinguish historical information from invented dialogue. Do not invent dates when the reference material is silent.
This is an editorial starting workflow, not a measured performance result.
Find the right plan for you
Free software
Explore official plans ↗Free software Models, hardware and hosting are separate
SillyTavern is free open-source software under AGPL-3.0. The interface requires an inference backend: local options use the user's hardware, while hosted model APIs can charge according to their own plans. Installing the frontend does not create free credits for those providers.
What is free and what can cost extra
SillyTavern is free open-source software under AGPL-3.0. The interface requires an inference backend: local options use the user's hardware, while hosted model APIs can charge according to their own plans. Installing the frontend does not create free credits for those providers.
Optional voice, image, search and other integrations can add costs or setup requirements. Evaluate the complete configuration, including model storage and maintenance, rather than treating the word free as a guarantee that every connected capability has zero operating cost.
A few things worth knowing
Frequently asked questions
Is SillyTavern free?
Yes. It is open-source software under AGPL-3.0.
Does it include a language model?
No. A separately configured inference backend produces the responses.
Can I use local models?
Supported local API routes are documented, with hardware and model requirements of their own.
Are character cards only for fictional characters?
No. They can also describe a scenario or task-specific assistant role.
Does World Info always include every fact?
No. Entries are activated and budgeted according to the configured rules.
Are cloud model calls free?
Not necessarily. Each provider has separate pricing and allowances.
Is it a simple hosted chatbot?
It is primarily a configurable local frontend with a learning curve and optional integrations.
The details behind your decision
Explore platforms, inputs and outputs, licensing, and access requirements.
Four useful starting projects
Four useful starting projects
- Develop an interactive story: Create a character card and a small set of setting facts, then test a short scene. Check whether the character follows the intended constraints before adding a large lore collection or several interacting roles.
- Practice a role-based conversation: Build a fictional interview or language-practice scenario with clear goals. Ask the model to stay within the exercise and review its feedback against the actual learning objective.
- Create a specialized assistant: Define a role for reviewing a particular type of document or brainstorming within a domain. Keep factual reference material separate from personality instructions so the assistant's style does not obscure evidence requirements.
- Compare model behavior: Use the same card and input with different configured backends. Record the model and settings, and compare instruction following and useful output rather than judging only response length or apparent enthusiasm.
Platforms, inputs, outputs and setup
Platforms, inputs, outputs and setup
Current documentation describes a Node.js-based application and installation routes for Windows, Linux, macOS, Android through Termux and Docker. Platform instructions and backend requirements differ, so follow the route actually intended for the device.
World Info and character prompts influence the model but do not guarantee compliance or factual accuracy. Long histories also compete with reference material for context space. Keep essential instructions clear and inspect context behavior before relying on a complex setup.
Data, permissions and licensing
Data, permissions and licensing
Conversation data is sent to the configured inference backend as needed to generate responses. Local installation alone does not make a cloud-connected conversation private to the machine; inspect the chosen provider and any active extensions.
Treat imported character cards and extensions as third-party material. Review their content and code where appropriate, keep credentials out of shared files and check the AGPL-3.0 obligations before distributing or operating a modified version for others.
Sources, documentation and help
Official sources and review date
Official sources and review date
Reviewed October 3, 2026. Product facts come from the official sources below. Suggested projects, prompts and review methods are FindGoodAI editorial guidance, not measured performance results.
Your experience can help someone choose
COMMUNITY
Sign in to rate and comment
Tried it? Share your experience
Loading experiences…
Manage my comments
Community and moderation
Saved tools are private. Approved comments are public; edits return to moderation. Comment counts include approved comments and replies. Each person has one rating. New ratings are automatically approved; withdrawn, unapproved or invalid ratings do not count. Share real experiences and avoid spam, private information or personal attacks.
For moderation appeals or data requests: support@findgoodai.com
NEXT STEP
Explore SillyTavern ↗Get started with SillyTavern through its official guides
Bring a real task and see how it fits the way you work.
Jump to a section
User experiences