What is SillyTavern?

SillyTavern is a local web interface that lets you chat with AI characters using LLMs. You can import character cards that others have made, or create your own from scratch.

For more details, check out the official website.

If you're new to AI roleplay, you might want to get a taste for it, so to speak, without paying anything first. Try free sites like character.ai, janitorai.com, or chub.ai to see if you enjoy the format. Just keep in mind: using those freemium services is like eating at McDonald's. SillyTavern is like being your own chef — you eat exactly what you want.

Installation

SillyTavern is available on almost any device — Windows, macOS, Linux, and even Android.

Follow the official installation guide. I recommend using the Git installation method for all operating systems.

Setting Up Your First Character

Character cards are .png or .json files that contain a character's personality, appearance, and backstory. You can download one of my character cards or create your own from scratch. To download a card from this site, just right-click the image and choose Save image.

To add a character in SillyTavern, click the ID card button next to the smiley face icon. From there you can choose to import an existing card or build a new one.

Connecting to an API

Click the plus icon in SillyTavern to open the connection panel. This is where you link an AI provider so your characters can actually talk.

Here's the setup I use and recommend. For API, choose Chat Completion. For Chat Completion Source, choose DeepSeek. You'll need an API key from the DeepSeek Platform and you'll have to buy tokens — don't worry, it's very cheap. For the model, use deepseek-v4-pro.

Tips & Tricks

The parts that take the longest to dial in are your preset and your extensions. The preset tells the LLM how to roleplay, so it's important to get right. Here's what I use to save you the trial and error.

Preset: I use peepsqueek - DS R1 — you can grab it here along with my other presets. It's my go-to preset and works great with DeepSeek.

Message Limiter: This extension limits the number of messages sent back to the LLM. It's a lifesaver for not burning through your tokens too fast. I also recommend turning on automatic message summarization in SillyTavern to keep long roleplays coherent without bloating the context.

Guided Generations: This extension lets you steer the LLM's responses whenever you want, giving you direct control over how a reply plays out.