Sovereign AI & Dynamic Router Sandbox

Sovereign Edge LLM Sandbox & Prompt Router

Runs localized Large Language Models entirely inside your browser utilizing WebGPU acceleration, combined with an outbound text router helper.
The Github for this webpage is at https://github.com/hpssjellis/gemma4-lite-rt-lm
Active PWA for installation of this web page for offline use is at https://webmcu-ai.github.io/local-gemma4-pwa/index.html
By Jeremy Ellis Github, or LinkedIn: Use at your own risk. May 2026

Select Local Weights Model Target:

Activate LLM on page open

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

Exclude user prompt tracking from terminal window

Custom Helper Target URL: then ctrl-v to paste

*Highlight parts of the streaming text inside the active window above to instantly port context layers to external systems.

⚙️ System Instruction Rule Editor

Edit System Context Layer: *Note: Changes made here apply automatically to your next prompt stream. If your conversation loops, re-click "Compile & Instantiate Local Engine" to clear the context cache completely.

🌐 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...