Locally Uncensored vs SillyTavern
Both SillyTavern and Locally Uncensored are built for people who want unrestricted AI without cloud censorship. But they target different use cases.
SillyTavern is the king of AI roleplay and character chat — a powerful tool with a massive character card ecosystem. Locally Uncensored is a general-purpose local AI app that combines chat, image generation, and video creation in one interface.
Feature Comparison
| Feature | Locally Uncensored | SillyTavern |
|---|---|---|
| AI Chat | Yes | Yes |
| Character Cards | Card v1, v2 and v3 import as personas, plus 25 built-in personas | Thousands (community) |
| Image Generation | Yes (ComfyUI built-in) | Yes (via SD / ComfyUI extension) |
| Video Generation | Yes (Wan 2.1) | No |
| Coding Agent | Yes | No |
| MCP Tools | 13 built-in | No |
| File Upload + Vision | Yes | No |
| Thinking Mode | Yes | No |
| Multi-Provider | Ollama + OpenAI + Anthropic | API connections |
| Model Load/Unload | Yes (VRAM mgmt) | No |
| Granular Permissions | 7 categories, 3 levels | No |
| Uncensored | Yes | Yes |
| Setup Complexity | One command | Node.js + config |
| Group Chat / Multi-char | Yes (2 to 4 models, each with its own persona) | Yes |
| World Building Tools | No (a card's lorebook is not used) | Yes |
| Modern UI | Glassmorphism | Functional |
| License | AGPL-3.0 | AGPL |
Role Play in Locally Uncensored 3.0.7
Since version 3.0.7, Locally Uncensored covers the core of character chat itself.
- Character cards: import a card as a PNG or JSON file, in card version 1, 2 or 3. It becomes a persona with the card picture as its tile. A new chat with that persona opens with the card's greeting as the first answer, and the arrows switch between alternate greetings until you write. A card persona exports back as a version 2 JSON file. A lorebook inside a card imports with it but is not used.
- Group chats: two to four models answer in turn inside one conversation, each on its own persona. A member whose model has left the model list sits out until it comes back and shows as unavailable in the group menu, and each turn keeps the persona it was written as (issues 165 and 169).
- Sampling per persona and per group member: a persona can carry its own temperature, top P, top K, min P, max tokens, repeat, presence and frequency penalty and DRY multiplier. They apply wherever the persona speaks, also on its seat in a group chat. An empty field follows the chat, and values set in the chat's Sampling popup still win (issue 175).
- Branches and answer versions: Duplicate chat makes a full copy with persona, group members and sampling values. Branch from here starts a new chat with the history up to that message. Regenerate, and Save & resend on an edited message, keep the earlier answer as a version you can page back to with arrows. Save on an edited earlier message changes only its text and keeps every later turn (issues 174 and 172).
Where this stands: character cards were checked in the app with Holo and Qwen models, not with Gemma. Duplicate chat, Branch from here, the kept answers, sampling per persona and the group chat fixes have run in automated tests only so far. Duplicate chat is not offered in Remote Access.
Where Locally Uncensored Wins
- Image + Video Generation built-in — Stable Diffusion/FLUX images and Wan 2.1 video work out of the box. SillyTavern supports image generation through its Stable Diffusion / ComfyUI extension once configured, but video generation isn't part of its scope.
- Simpler setup — One command vs SillyTavern's multi-step Node.js setup with API configuration.
- Modern UI — Clean glassmorphism design with dark/light mode. SillyTavern's interface is functional but dated.
- AGPL-3.0 license — Same open source commitment as SillyTavern. Fork it, modify it, contribute back.
- All-in-one — Chat, images, video, model management — one app instead of multiple tools.
- Coding Agent — Built-in coding agent that reads your codebase, writes files, and runs commands. SillyTavern has no coding capabilities.
- 13 MCP Tools — Web search, file operations, shell access, screenshots, and more — all integrated as tool-use capabilities.
- File Upload with Vision — Drag and drop images into chat and vision models describe them. SillyTavern is text-focused.
- Thinking Mode — Collapsible reasoning blocks that work with all providers. Great for complex problem solving.
- Multi-Provider Support — Use Ollama, OpenAI, and Anthropic simultaneously. SillyTavern connects to APIs but not as a unified multi-provider system.
- Model Load/Unload — Manage VRAM directly from the UI by loading and unloading models on demand.
Where SillyTavern Wins
- Character ecosystem — Thousands of community-created character cards. SillyTavern IS the platform for AI roleplay.
- World building — Lorebooks, scenario descriptions, and persistent memory for complex narratives.
- Mature community — Years of development, extensive documentation, active Discord.
- Advanced prompting — Fine-grained control over system prompts, jailbreaks, and character behavior.
The Verdict
Choose SillyTavern if AI roleplay and character chat is your main use case. Its community card ecosystem, lorebooks and fine prompt control go further than anything in Locally Uncensored.
Choose Locally Uncensored if you want a general-purpose uncensored AI app that also generates images and video, has a built-in coding agent, 13 MCP tools, vision support, and multi-provider flexibility. It's the better choice if you need more than just chat, or if you want a cleaner, simpler experience.
Try Locally Uncensored
Free, open source, AGPL-3.0 licensed. One command to get started.
View on GitHubNo GPU? LU Cloud runs in a browser, paid from credit packs that start at 15 EUR and need no subscription. See what it costs.
More comparisons
- Alternatives and comparisons, every head to head on one page
- Locally Uncensored vs LM Studio, the most asked comparison
- Locally Uncensored vs Ollama, app against engine
- What Locally Uncensored is, what runs on your own machine and what runs on LU Cloud