What is this?
WEBWAIFU 3 is a browser-based VRM avatar companion powered by local or cloud AI. Chat with your 3D character using text or voice, with real-time lip sync and animations.
Before You Start
- BYOK (Bring Your Own Key) — WEBWAIFU 3 does not include hosted cloud credits or bundled API keys. For cloud chat/TTS, you must use your own provider key(s).
- LLM Required — You need either a local model server (Ollama or LM Studio) or your own cloud API key (OpenAI / OpenRouter).
- Ollama Network Access — In Ollama's settings, enable "Allow through network" so the browser can reach it. You also need to set allowed origins: set the environment variable
OLLAMA_ORIGINS=* (or add your app's URL) before running ollama serve. On Windows, set it in System Environment Variables and restart Ollama. - LM Studio CORS — Enable CORS in the LM Studio server settings.
- API Keys are Stored Locally — All API keys (LLM, TTS) are saved in your browser's IndexedDB. They never leave your machine except when sent directly to the provider APIs.
TTS (Text-to-Speech)
- Kokoro — Runs 100% locally in your browser via WebGPU/WASM. 28 voices, no API key needed.
- Fish Audio — Cloud TTS with custom voice cloning. Requires API key.
Quick Setup (Free)
- Install Ollama or LM Studio
- Pull a model:
ollama pull llama3.2 - In Ollama settings, enable "Allow through network"
- Set
OLLAMA_ORIGINS=* as an environment variable and restart Ollama - Open settings (gear icon) → AI tab → Refresh Models
- Start chatting!