LM Studio Model Not Showing Up? The Folder Structure Nobody Explains (2026)

You downloaded a GGUF file from Hugging Face, dropped it into LM Studio's models folder, opened the app — and nothing. No error. No warning. The model just doesn't exist as far as LM Studio is concerned. You check the file is actually there. It is. You re-download it, just in case. Still nothing.

This isn't a bug you're causing. It's an undocumented rule about where exactly inside the models folder a file needs to live, and once you know it, the fix takes ten seconds.

The Actual Rule Nobody Tells You

LM Studio does not scan its models folder flatly. It only detects a GGUF file if it sits inside a specific two-level folder structure that mirrors how models are organized on Hugging Face:

~/.lmstudio/models/
└── publisher/
    └── model-name/
        └── model-file.gguf

So if you have a model called ocelot-v1 published by infra-ai, LM Studio expects:

~/.lmstudio/models/
└── infra-ai/
    └── ocelot-v1/
        └── ocelot-v1-instruct-q4_0.gguf

Drop the same file directly into ~/.lmstudio/models/ with no subfolders, and LM Studio silently ignores it. No error dialog, no log entry, nothing in the documentation that flags this at the point of failure — this exact behavior has been reported as a rough edge by multiple users independently, and the underlying cause is always the same missing subfolder layer.

On Windows, your path is C:\Users\[username]\.lmstudio\models\. On macOS and Linux, it's ~/.lmstudio/models/ (older installs may still use ~/.cache/lm-studio/models/).

The Fix: Two Ways to Do It Right

Option 1 — Use the lms import command (recommended)

LM Studio ships a CLI specifically for this. If you already have lms set up (see our GPU troubleshooting guide if not), importing a stray GGUF file is one line:

lms import <path/to/model.gguf>

Follow the interactive prompt — it asks you to confirm the publisher/model naming, then places the file in the correct nested folder automatically. This is the safest option because you can't get the folder structure wrong.

Option 2 — Fix the folder structure manually

If you'd rather not use the CLI, just recreate the structure yourself:

1. Inside your models directory, create a folder named after the model's publisher (this can be anything reasonably descriptive — it doesn't have to exactly match Hugging Face).
2. Inside that, create a second folder named after the specific model.
3. Move your .gguf file into that innermost folder.
4. Restart LM Studio, or open the My Models tab and it should refresh automatically.

That's the entire fix. No reinstall, no re-download — just two folders.

Finding Your Actual Models Directory

If you're not sure where your models folder even lives (especially after a custom install location), open LM Studio and go to the My Models tab — there's a setting there that shows and lets you manage your current models directory path directly, rather than guessing at OS-default paths.

You can also point LM Studio at an entirely different drive or folder from that same screen, which is worth doing early if you're planning to download a lot of larger models and don't want them eating your primary drive.

Bonus: Already Have Models Downloaded in Ollama?

If you've been using Ollama and want to try the same models in LM Studio without re-downloading multi-gigabyte files twice, there's a community tool called StudioLink built specifically for this. It scans Ollama's model manifests, locates the underlying GGUF blobs, and imports them into LM Studio's expected folder structure using hard links by default — so you're not duplicating disk space for models you already have.

It's a third-party tool, not an official LM Studio feature, but it solves exactly the cross-tool folder-structure headache this post is about, in reverse.

Cheat Sheet

Final Advice

Before you assume a redownload or a fresh install will fix a model that "isn't there," check the folder structure first — this single undocumented rule accounts for a surprising share of "LM Studio isn't detecting my model" reports, and it costs nothing to rule out. lms import is the safer long-term habit since it removes the guesswork entirely.

Once your models are actually showing up, our guide to fixing "Failed to load model" errors covers the next thing that tends to go wrong, and if you're deciding which model is even worth downloading in the first place, our LM Studio Bionic guide breaks down picks by hardware tier.