Legacy Lantern
Local & private

A light for the stories we carry

Some memories deserve
their own glow.

Create a remembrance from the details only you know. Open-weight AI runs on your device, so family stories never leave it.

Light a lantern ↓
01No accountBegin without a login.
02No uploadInference happens here.
03No lock-inExport your keepsake.

Remembrance studio

Tell the machine what
the machine could never know.

0/900

Model downloads once, then stays in your browser cache.

The ghost in this machine is open

Private by architecture,
not by promise.

Legacy Lantern uses Qwen2.5-0.5B-Instruct through the open-source Transformers.js runtime. Model weights execute locally with WebGPU or WASM. After the first model download, your memories can stay off the internet—and outside anyone else’s database.