v2.3.1 just dropped — small version number, big quality-of-life improvements.
in-app Ollama install
no more "go to ollama.com, download, install, restart". click Install Ollama in the onboarding wizard, watch the progress bar (with real download speed and elapsed time), done. silent install, auto-start, auto-detect. zero manual steps.
this is what "plug & play" should feel like.
configurable ComfyUI port & path
this one's a direct response to user feedback. ComfyUI recently moved from a browser-based interface to a desktop app — which uses a different port than the old default 8188.
result: LU couldn't find ComfyUI even though it was running. multiple users reported this in our GitHub Discussions within hours of v2.3.0.
v2.3.1 fix: the ComfyUI port is now fully configurable in Settings > ComfyUI (Image & Video). same for the path — editable with a Connect button. no need to go through onboarding again.
previously the port was hardcoded in 20+ places across the codebase. now it's a single config value.
ComfyUI install progress (actually works now)
the one-click ComfyUI install now shows real step-by-step progress:
- Step 1/3: Clone repository
- Step 2/3: Install PyTorch
- Step 3/3: Install dependencies
with an elapsed timer. before this, the install screen got stuck at "Starting..." forever because the Rust backend thread never reported completion back to the frontend.
provider status that actually means something
the connection dots next to providers in Settings now show actual status:
- 🟢 green = connected and responding
- 🔴 red = connection failed
- ⚪ gray = not checked yet
before this they were always green. which is worse than showing nothing at all.
the bigger picture
v2.3.0 (yesterday) was the big feature release — ComfyUI plug & play, 20 model bundles, image-to-image, image-to-video on 6 GB VRAM.
v2.3.1 is the "we actually listen to our users" release. someone reported ComfyUI Desktop not connecting at 11pm, the fix shipped by noon next day.
if you tried v2.3.0 and ComfyUI didn't work — update and check Settings. the port config should fix it.
GitHub: PurpleDoubleD/locally-uncensored
Download: v2.3.1 Release
Top comments (0)