The all-in-one local AI studio for your desktop: chat, image and video generation and a coding agent in one free, open source app. Windows and Linux. No Docker, no terminal, no cloud required.
TypeScript agentai-chatai-studioall-in-onecoding-agentdesktop-appfluximage-generationlinuxllmlocal-ailocal-llmopen-sourceprivacyself-hostedstable-diffusionuncensoreduncensored-aivideo-generationwindows
/**
* Read-only questions about the model folders LU does NOT own.
*
* Deliberately separate from `detectProviderModelPath` in discover.ts: that
* one asks the backend for a DOWNLOAD TARGET and the backend creates the
* folder to hand one back. Storage asks a different question ("where is
* it, and is it there at all") and must not leave a folder behind on a machine
* that has no LM Studio.
*/
import { backendCall } from './backend'
import type { LmStudioModelDir } from '../lib/model-storage-rows'
/**
* LM Studio's models folder, without creating anything.
*
* Any failure (web build, bridge without the command, older backend) answers
* "not installed, no path", which is what the panel should say when it cannot
* find out: the row is informational and must never block Settings.
*/
export async function lmStudioModelDir(): Promise<LmStudioModelDir> {
try {
const res = await backendCall<{ installed?: boolean; path?: string | null }>('lmstudio_model_dir')
return { installed: res?.installed === true, path: res?.path ?? null }
} catch {
return { installed: false, path: null }
}
}