๐ [GUIDE] BOTBOORU FRONTENDS & AI CHAT CLIENTS (2026)
BotBooru is a character card repository; it does not process chat messages or run AI models itself. To chat, you need:
- Frontend (UI): The interface where you import cards and interact with the character.
- Backend (LLM): The engine that generates responses. This can be run locally on your own hardware (e.g., KoboldCpp) or remotely via a cloud API (e.g., OpenRouter, Claude, OpenAI).
๐ QUICK-SELECT MATRIX
| Frontend | Platform | Setup | Backend Access (Local / Paid API / Free Option) | Status / Trust Level |
|---|---|---|---|---|
| SillyTavern | PC, Mac, Android | Moderate | Local / Paid API / Free (Horde integration) | High (Industry Standard) |
| Front Porch AI | PC, Mac, Linux | Low | Local (Auto-installer) / Paid API | High (Active Desktop App) |
| Agnai | Web, Mobile | Low | Local / Paid API (or Native Sub) / Free (Site Models) | High (Established Web App) |
| Koboldcpp | PC, Linux | Moderate | Local Only (GGUF runner) | High (Direct Backend UI) |
| Saucepan | Web | Low | Local / Paid API / Free (Unlimited basic + daily premium) | Active (Alternative Web UI) |
| HammerAI | PC, Web, Android | Low | Local (WebGPU) / Paid API / Free (Context-limited cloud) | Active |
| Tavo | Android, iOS | Low | Paid API / Custom Proxies | Active Mobile Client |
| ChatterUI | Android | Moderate | Local (On-device GGUF) / Paid API | Active Mobile Client |
| Marinara | PC, Web | Low | Local / Paid API | Niche / Unorthodox RPG |
| Serene Pub | PC, Mac, Linux | Low | Local / Paid API | Active Alpha (Streamlined group chat focus) |
| Pyre | Web, Self-Hosted | Low-Mod | Local / Paid API | Experimental (Very New) |
๐ ESTABLISHED & POWER-USER FRONTENDS
SillyTavern
[Stable] [Hybrid] | PC / Mac / Android (Termux)
- Setup Difficulty: Moderate (Requires Node.js environment)
- Card Compatibility: Full V2 PNG/JSON spec.
- Backend Access:
- Local: Fully supported (Ollama, KoboldCpp, Text-Gen-WebUI, GGUF, etc.).
- Paid API / Proxy: Fully supported (OpenRouter, Claude/Anthropic, OpenAI, custom reverse proxies, and custom endpoints).
- Free Option: No built-in free cloud models, but has native integration for the community-run, crowdsourced AI Horde.
- Pros: The community standard for custom roleplay. Includes extensive lorebook support, expression sprites, custom CSS, scripting, and extensions (TTS, image generation).
- Cons: Steep learning curve. Manual command-line installation and updates are required on mobile.
- Links:
Front Porch AI
[Stable] [Hybrid] | Windows / Mac / Linux (Native)
- Setup Difficulty: Low (Standard desktop installer)
- Card Compatibility: High (V2 spec, native Chub.ai browser, and parses legacy Backyard AI
.byafcards). - Backend Access:
- Local: Native support. Features an automated internal KoboldCpp downloader to run GGUF models out-of-the-box on your GPU/CPU, alongside external local APIs.
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI, and custom endpoints).
- Free Option: None. You must provide your own hardware or API key.
- Pros: No complex developer dependencies. Includes built-in local RAG memory, text-to-speech (Kokoro/Piper), and speech-to-text (Whisper). Tracks persistent state details (trust levels, narrative objectives, emotions).
- Cons: Less extensible than SillyTavern; smaller plugin ecosystem.
- Links:
๐ BROWSER-BASED & CROSS-DEVICE FRONTENDS
Agnai
[Stable] [Hybrid] | Web Browser / Mobile
- Setup Difficulty: Low (No installation required)
- Card Compatibility: High (Import cards directly and sync them across devices via account).
- Backend Access:
- Local: Supported (can connect to local endpoints like KoboldCpp or Ollama via browser settings).
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI, custom endpoints). Agnai also offers its own paid subscription tiers ("Aesthete" and "Artificer") to natively run premium cloud models such as DeepSeek 3.2, Qwen3, and Magnum without needing an external API key.
- Free Option: Yes. Directly hosts several free, zero-config models (including Gemmasutra 9B, Llama 3.1 8B, and Mistral Nemo 12B) with an 8K context limit and a 400-token generation limit. No external accounts or API keys required.
- Pros: Instant setup. Offers zero-install cross-device synchronization and highly capable native free hosting, making it highly accessible for low-spec or budget-restricted users.
- Cons: Dependent on remote web hosting or self-hosting stability.
- Links:
Saucepan
[Active] [Hybrid] | Web Browser
- Setup Difficulty: Low (No installation required)
- Card Compatibility: Moderate.
- Backend Access:
- Local: Supported (Ollama, KoboldCpp).
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI, custom endpoints).
- Free Option: Yes. Offers unlimited free messages using their basic "Arita" model, plus 20 daily free messages of higher-tier models.
- Pros: Clean, minimalist, distraction-free layout. Ideal for low-end hardware users requiring free access.
- Cons: Highly minimalist; lacks deep customizable settings or lorebook management compared to dedicated desktop clients.
๐ฅ๏ธ LOCAL-FIRST FRONTENDS
KoboldCpp / Kobold Lite
[Stable] [Local-First] | Windows / Linux
- Setup Difficulty: Moderate (Single
.exewith basic settings menu) - Card Compatibility: High (Standard card loader).
- Backend Access:
- Local: Specifically designed as a local GGUF model runner. Kobold Lite is the built-in browser UI served directly by your local backend.
- Paid API / Proxy: No. It connects exclusively to your running KoboldCpp backend.
- Free Option: Free to run on your own hardware. No external free cloud backends are supported.
- Pros: Extremely lightweight, fully private, and runs entirely in your browser with zero installation. Highly optimized engine for local GPUs.
- Cons: Visually basic, barebones UI designed for pure text input; lacks high-end customization options.
HammerAI
[Active] [Hybrid / Local-First] | PC / Web (WebGPU) / Android
- Setup Difficulty: Low (Click-to-run desktop executable or native mobile APK)
- Card Compatibility: Moderate.
- Backend Access:
- Local: Supported. Runs models directly in your browser or desktop app via WebGPU using your own hardware (no Nvidia CUDA toolkit installation required).
- Paid API / Proxy: Supported (allows you to bring your own OpenRouter, Claude, or OpenAI keys).
- Free Option: Yes. Offers unlimited cloud-hosted messaging on their default models.
- Pros: Zero-setup local execution using WebGPU.
- Cons: The free-tier cloud models have heavily restricted context sizes, meaning the AI will forget the beginning of the conversation sooner than paid API alternatives.
๐ฑ MOBILE-SPECIFIC FRONTENDS
Tavo
[Active] [Hybrid] | Android / iOS
- Setup Difficulty: Low (Standard App/Play Store install)
- Backend Access:
- Local: No. Cannot host local models on the phone.
- Paid API / Proxy: Supported (OpenRouter, OpenAI, and custom proxies).
- Free Option: No. You must provide a paid API key or connect to a custom proxy.
- Pros: Clean, mobile-optimized alternative to setting up Termux. Includes built-in API setup guides.
- Cons: Relies purely on remote connections.
ChatterUI
[Active] [Hybrid] | Android (Sideloaded APK)
- Setup Difficulty: Moderate (Requires manual APK installation and file navigation)
- Backend Access:
- Local: Supported. Can load and run smaller GGUF models directly on your phone's storage and hardware.
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI, custom proxies).
- Free Option: No.
- Pros: Open-source, highly lightweight, and capable of fully offline mobile roleplay.
- Cons: Sideloading required; UI is functional but unpolished.
๐งช NEW & EXPERIMENTAL FRONTENDS
Marinara Engine
[Active] [Hybrid] | PC / Web
- Setup Difficulty: Low (One-click installer)
- Backend Access:
- Local: Supported (Ollama, KoboldCpp, local GGUF).
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI).
- Free Option: No.
- Pros: Geared heavily toward "agentic" RPG setups. Uses background system agents to handle character inventories, live tabletop stat updates, and automatic scene/room illustrations.
- Cons: Niche, non-traditional roleplay interface; can feel cluttered if you only want a standard visual novel/chat layout.
Serene Pub
[Alpha/Active] [Hybrid] | PC / Mac / Linux
- Setup Difficulty: Low (Download and run executable)
- Card Compatibility: Moderate.
- Backend Access:
- Local: Supported (Ollama, KoboldCpp).
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI, and custom proxies).
- Free Option: None.
- Pros: Designed specifically as a streamlined, user-friendly desktop alternative to SillyTavern. It features a highly capable, zero-clutter Lorebook implementation and robust group chat settings. Uniquely supports multi-user cooperative roleplay, allowing you and multiple real-life friends to connect to the same chat session and take turns roleplaying with characters and the AI simultaneously.
- Cons: Still in active Alpha development; some advanced modular customization features are still missing, and you may encounter baseline bugs.
- Links:
Pyre
[Experimental] [Hybrid] | Web Browser / Self-Hosted
- Setup Difficulty: Low (Web app) to Moderate (Self-hosting via GitHub)
- Backend Access:
- Local: Supported (Ollama, KoboldCpp).
- Paid API / Proxy: Supported (OpenRouter, Claude, OpenAI, custom proxies).
- Free Option: No.
- Pros: Highly lightweight, fast, and modern open-source web interface. Fully self-hostable.
- Cons: Still in early development; currently less stable than established alternatives and prone to bugs or missing baseline features.
- Links: