Skip to content
ZushZush Docs

LM Studio Setup for Local AI

LM Studio mode connects Zush to the local OpenAI-compatible server managed by LM Studio. Use it when you want to choose your own vision-capable model and control model loading in LM Studio.

LM Studio model and host settings in Zush for Mac.
The Mac interface shown here connects Zush to an installed vision-capable model. Windows provides the same LM Studio mode in Zush Settings.

Requirements

Section titled “Requirements”
  • the latest Zush version for Mac or Windows
  • LM Studio installed
  • a downloaded vision-capable model
  • the LM Studio local server running

Set up LM Studio

Section titled “Set up LM Studio”
  1. Install and open LM Studio.
  2. Download a vision-capable model in LM Studio.
  3. Open Developer in LM Studio and turn on Start Server.
  4. Open Zush Settings and select AI Modes.
  5. Select LM Studio.
  6. Confirm the host. The default local endpoint is http://127.0.0.1:1234.
  7. Refresh the model list and select the installed vision-capable model.
  8. Run Test and wait for a successful connection.
  9. Test a small batch in AI Rename before using folder monitoring.

Troubleshooting

Section titled “Troubleshooting”
ProblemWhat to check
LM Studio is not detectedInstall and open LM Studio, then return to AI Modes
Server is not runningIn LM Studio, open Developer and enable Start Server
No models appearDownload and load a vision-capable model, then refresh in Zush
Connection test failsConfirm the host is http://127.0.0.1:1234 unless you intentionally changed it
Model is too slowLoad a smaller vision model or compare the Ollama setup

Privacy

Section titled “Privacy”

With the default loopback endpoint, supported analysis runs through LM Studio on your Mac or Windows PC. Zush does not silently switch to cloud analysis while LM Studio mode is selected.

Section titled “Related docs”