Skip to main content

🤖 STELLAR AI

Private browser-local space chat and an offline educational demo

Conversations stay in browser storage. No model request occurs until you explicitly load Bonsai.

💬 Chat History

🌟 Stellar AI

BROWSER-LOCAL AI

Run Bonsai privately with WebGPU

Model weights download only after you press Load model. Prompts and replies stay in this browser; no inference service receives them.

WebGPU check pending. No model has been downloaded.
Compatibility, storage, and source

Requires a WebGPU-capable browser and GPU. The 27B option is the requested approximately 4 GB experiment and needs substantially more working memory than its download size. Models are cached locally and can be removed here.

Powered by the MIT-licensed BitGPU Bonsai experiment, pinned to version 0.19.1.

🌌

Welcome to Stellar AI

Load Bonsai for private WebGPU inference or use the clearly labelled offline demo.

💬Natural Conversations
🖼️Image Analysis
📚Knowledge Base
💾Save History

Try asking:

0 / 4000
💻

Run Stellar AI Locally

Download the Stellar AI CLI to run AI chat from your terminal with advanced tool calling, file system integration, and multi-agent support.

🎙️ LiveKit status

LiveKit is not configured in this static deployment. A future voice service must use a free, separately reviewed token issuer and an explicit microphone permission flow. The browser-local Bonsai and speech-to-text controls above remain available without that service.