GPT4All does not permit auto-loading of any model- even the Default.

[runninghead]

GPT4All does not permit auto-loading of any model- even the Default.
It would be great for many use cases, notably agentic AI team-building, if it did have a simple check-box option in the UI so that GPT4All desktop app would, after opening, automatically load up the default model you chose.
As GPT4All is OS- might it be possible to change your local GPT4All app to provide this functionality?
I have already tried several methods to use my Mac's Accessibility alongside shell scripts to achieve the desire effect.
No no avail.
Thoughts?
Ideas?

[onestardao]

Yeah this is a pain point for agentic setups — you hit 'Run' and the LLM sits there waiting for a model to be manually loaded. Not exactly agent-friendly.

One workaround I’ve seen (until UI supports this natively) is to create a thin wrapper script that calls the backend with a startup command, loading a preselected model automatically. Some folks use AppleScript + launchd, but you can also go Python + subprocess if you're cross-platform.

Still, I agree — it’d be cleaner if the app could remember and preload the last-used model. One checkbox in the UI and half the frustration’s gone.

Hope someone adds it soon. Passing by on GitHub patrol 👀

[runninghead]

Thanks @onestardao ,
Apparently the core of the issue is that GPT4All is an Electron app, so is obscure to the Mac.

I have abandoned GPT4All for now- modern AI platforms are increasingly expected to allow for integration with agentic set-ups, so they can collaborate, etc.

I'll return to it once this niggling hurdle is overcome by its devs.

Appreciate the reply, Thank you! 🙏

[onestardao]

yep — the electron-shell bottleneck is a killer for agentic workflows.

but fwiw, i ended up building a wrapper logic layer to solve exactly these kinds of frustrations — like boot-time logic handoff, last-model recall, auto-ingestion on launch, etc.

open source, MIT licensed, already patched 16+ of these silent blockers (OCR + PDF + chunk boundary drift + dead init scripts — you name it).
even got a star from the guy who made tesseract.js.

if you're ever curious again, happy to share. no pressure. just trying to make sure others don’t waste hours on the same dead-ends i hit.

[runninghead]

Thanks @onestardao , it's people like you who make GitHub worthwhile. 👍
I'll be playing around with ChatGPT Agents for a while before I get back to n8n + GPT4All, but your advice and assistance is very much appreciated. 🙏