Analysis updated 2026-05-18
Chat with a fully local, private AI avatar using in-browser language, speech, and TTS models.
Connect your own API keys to providers like OpenAI, Anthropic, or Groq to power a talking avatar.
Build and customize a personal 3D avatar character with your own VRM model, name, and personality.
View the avatar on a Looking Glass holographic display or through a WebXR headset.
| looking-glass/liteforms-web | 0xkinno/neuralvault | 0xmayurrr/ai-contractauditor | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | TypeScript | TypeScript | TypeScript |
| Setup difficulty | easy | hard | easy |
| Complexity | 2/5 | 4/5 | 2/5 |
| Audience | developer | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Local models may take a while to download on first run depending on hardware and network speed.
Liteforms Web is a browser-based chat app built around a three-dimensional animated avatar instead of a plain text interface. The avatar is rendered in real time using the VRM format, a common standard for 3D character models, and it responds with facial expressions, mouth movement, and animation as the conversation happens. You can customize the character's name, pronouns, and personality, and upload your own VRM model through a built-in editor. The app is built to work without any account or hosted service. You can run it locally and choose to use language models, speech recognition, and text-to-speech entirely inside the browser, using ONNX and Transformers-based models, Kokoro for text-to-speech, and Distil-Whisper for speech recognition, all without sending data anywhere. If you prefer, you can instead connect your own API keys to external providers. Supported language model providers include OpenAI-compatible APIs, Anthropic, Google AI Studio, xAI, Mistral, Cerebras, NVIDIA, OpenRouter, Groq, and Ollama, among others. Supported speech providers include ElevenLabs, Deepgram, Azure Speech, Microsoft Edge TTS, and Google, among others. All of this is configured inside the app's own settings panel rather than through a server-side account. Under the hood, the project is built with Next.js, React, and TypeScript, with Three.js rendering the 3D avatar scene. It also supports Looking Glass holographic displays and WebXR headsets. Running it locally needs Node.js 20 or newer and npm, the first run may take a while since browser-local models need to download. Any API keys or credentials you enter are stored only in your browser's local storage and are never committed to the repository or sent to a project server. The project itself is open source, though the README notes that some libraries it depends on, such as the Looking Glass WebXR library, are proprietary rather than open source.
A browser-based chat app where you talk to a customizable, animated 3D VRM avatar, using either local browser AI models or your own API keys.
Mainly TypeScript. The stack also includes TypeScript, Next.js, React.
The project is open source, but some dependencies such as the Looking Glass WebXR library are proprietary.
Setup difficulty is rated easy, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.