Sovereign AI - On-Device Chat Sandbox

Local Edge LLM Sandbox (LiteRT-LM)

This page runs a localized Large Language Model entirely within your browser using WebGPU acceleration.
No cloud API keys required, no external server reporting.
The Github for this webpage is at https://github.com/hpssjellis/gemma4-lite-rt-lm
By Jeremy Ellis: Use at your own risk. May 2026

Select Local Weights Model Target:

Max Token Sequence Allocation:

Pipeline State: Uninitialized

[System Log]: Engine uninitialized. Select a model target and compile to load hardware kernels...

Enable active terminal window auto-scrolling during streaming

🌐 HTML Viewer

∑ LATEX Math Viewer

Auto-render calculations while generation streams

Rendered mathematics container...

🖼 Image Viewer (SVG & Inline Base64)

Auto-extract vector images during stream

Vector elements and inline image configurations render here...