Eviory LogoEviory
Back to Help Center

Troubleshooting LM Studio Local Server Connection

August 17, 2026 (1w ago)

LM Studio allows you to run models (such as Llama 3, Qwen 2.5, or Mistral) locally on your Mac or PC with GPU acceleration. When connecting Eviory OS, you must start the local developer server and enable Cross-Origin Resource Sharing (CORS).

3-Step Setup & Fix

Step 1: Start the Local Server in LM Studio

  1. Open LM Studio and download your desired GGUF model.
  2. Click on the Developer / Local Server icon (the <-> icon on the left navigation bar).
  3. Select your model at the top dropdown to load it into RAM/VRAM.
  4. Click the green Start Server button. By default, it runs at http://localhost:1234.

Step 2: Enable CORS Support

  1. On the Local Server configuration panel on the right sidebar, scroll down to Server Configuration.
  2. Make sure Enable CORS is toggled ON (true).
  3. Ensure Request Origin allows * or includes your Eviory domain URL.

Step 3: Connect in Eviory OS

  1. In Eviory OS, open Settings → AI Integrations.
  2. Select LM Studio / Local OpenAI Compatible.
  3. Set Endpoint URL to http://localhost:1234/v1 and save.

Troubleshooting Common Glitches

  • Model Crashes or Slow Responses: Check that your context length in LM Studio does not exceed your available GPU VRAM (e.g. adjust context to 4096 or 8192 tokens).
  • Port Conflict: If port 1234 is already used by another process, change the port in LM Studio to 1235 and update the URL in Eviory OS Settings accordingly (http://localhost:1235/v1).