Troubleshooting

App will not install or open

  • Re-download the installer from https://inference.js.org.

  • Confirm you picked the correct OS and architecture build.

  • On macOS, if blocked by security settings, open from System Settings after first attempt.

  • On Linux, make sure the file is executable if using AppImage.

Models are not loading

Checks:

  • Ensure Ollama is installed/running if using local host.

  • Verify remote host URL configuration in UI host selector.

  • Confirm proxy host is reachable if using port 52458 workflows.

Actions:

  • Restart app and retry model list in marketplace/chat.

  • Try another model to isolate whether one model is broken.

  • If using remote host, verify the host URL is still online.

Chat responses stop streaming

Possible causes:

  • Active stream was aborted.

  • Model backend became unavailable.

  • Tool call failed due to network/service issues.

Actions:

  • Start a new chat session and retry.

  • Switch to a known-good model/host.

  • Disable optional tools and retry prompt.

Auth issues

If sign-in state is inconsistent:

  • Open auth page and complete login again.

  • Make sure your browser completed the login redirect flow.

  • Sign out and sign back in from Settings/Auth screens.

Sync does not update

  • Confirm sign-in state is valid.

  • Enable sync toggle in Settings.

  • Check your internet connection and retry.

Still stuck?

  1. Restart the app.

  2. Try the latest standard build for your OS.

  3. Open an issue with steps to reproduce on https://github.com/sharktide/inferenceport-ai/issues.