Runs in your browser · No install · No cloud

A personal AI that remembers you
— entirely on your device.

DiciZ runs a real language model directly on your GPU via WebGPU, with a persistent on-device memory: tell it about you once and it remembers — across sessions, even after a reboot. Fully private — nothing is sent anywhere, and it works offline after the first load.

Verified end-to-end · real GPU compute · no API key · your data never leaves your machine
🧠

An AI that remembers you

A genuine language model runs in a browser tab and answers correctly — and it keeps a private memory of what you tell it, stored on your device. Pick a preset in the chat or paste any GGUF model URL; the first load downloads the model, then it is cached offline.

💬 Open AI Chat
remembers across sessionsoffline after first load100% private
🎬

AI frame generation

RIFE runs on your GPU via WebGPU and generates in-between frames to multiply FPS (2× / 4× / 8×), with motion-gating. Try the live demo or load your own clip — your video never leaves your machine, and the built-in benchmark measures the real gain on your clip.

🎮 Try frame generation
runs on your GPUno uploadhonest quality benchmark

How it works — and what's honest about it

On your GPU. Everything runs locally via WebGPU. Speed depends on your GPU; the AI is correct (engine-verified), not a mock.
It remembers — privately. What you tell it is embedded and stored on your device (IndexedDB), then retrieved when relevant. It survives refresh and reboot, and never leaves your machine.
Private by design. No cloud, no API key, no telemetry. Once cached, you can pull the plug and it keeps working.
Requires a modern browser with WebGPU (Chrome, Edge, recent Safari/Firefox). Best on a discrete GPU.