Features

Four studios, one window, zero cloud.

LU Labs auto-detects your local backends and runs text, code, image and video generation — private, offline and fast.

Chat with any local model.

Talk to any model you can run. Switch between them mid-conversation, watch the reasoning in thinking mode, and ground answers in your own documents with RAG.

  • Thinking mode with visible reasoning
  • RAG over local files & folders
  • Voice in, voice out — fully offline
Qwen 3.6 · 14BLoaded
Draft a cold email to a Series A founder.
On it — here's a three-line version that leads with the metric and keeps the ask soft…
62 tok/s
Codex · agentLoaded
Fix the failing auth test in verify().
Found it — verify() never checked claims.exp. Patched it and re-ran the suite…
2 edits · 0 failing

A coding agent that works on your PC.

Codex-style agent mode plans, edits and runs real tools across your files — locally, so your source never touches someone else's server.

  • Multi-file edits with diffs you approve
  • Runs commands & tests in a sandbox
  • Your code never leaves the machine

Image generation through ComfyUI.

Drive FLUX.1 and any ComfyUI graph from a clean prompt bar. Full resolution, no watermark, no per-image billing — it's your GPU.

  • FLUX.1 Schnell & custom checkpoints
  • Unlimited generations, no watermark
  • ComfyUI process auto-killed on close
IMGiridescent violet monolith, studio light1024²
LIVE
VID
00:04 / 00:06

Text to video on your own GPU.

Prompt a clip and let your hardware render it — no queue, no credits, no upload of your footage to anyone. Same ComfyUI backend, wired for motion.

  • Local render — no cloud queue
  • Your footage never gets uploaded
  • Preview and re-roll instantly

Ready to run it all locally?

Free to self-host, or start a hosted plan in seconds.