[Issue] (GUI/Studio) No way to use imatrix file converting "Local Model" to/via "GGUF/Llama.cpp"

[belfon]

---

name: Unsloth Studio Bug/Issue
about: Report a problem with the Unsloth Studio desktop app or web UI
title: No way to use imatrix file converting "Local Model" to/via "GGUF/Llama.cpp"
labels: issue, missing important functionality
assignees: ""

Environment

Where are you using Unsloth?

• [ X] Studio desktop application
• Studio web UI (unsloth studio)
• Unsloth CLI
• Python package or notebook
• Colab or Kaggle

Operating system and version:
Windows 11 10.0.106200

GPU model(s) and accelerator backend:

RTX 4080 CUDA

Versions:
Unsloth Version
v0.1.801-beta
Package Version
2026.8.19
Desktop App Version
0.1.801-beta
llama.cpp Version
b10472-mix-4b653db

What happened?

"Convert Local File" is a wonderful ability to have (and what made me download the software in the first place) but there is no functionality/method that allows the User to use a local imatrix file to create an imatrix GGUF. The software searches HF for imatrix file and errors. My "local file" is a model created via mergekit. I have a local imatrix file. Nothing in the UI or anywhere allows me to utilize the imatrix file I have.

Steps to reproduce:

1. Have a local file
2. Try to convert it to GGUF imatrix
3. Put a DAT/GGUF imatrix file, correctly named, in the originating model's folder. Unsloth Studio does not accept. And there is nowhere in the UI to upload it or anything like that. Tried imatrix.gguf, imatrix_unsloth.gguf, imatrix_unsloth.gguf_file

Says I can put it in /path/to/ but I'm not sure where that is other than the source model folder???

Expected behavior:
Let me use an imatrix file.

Actual behavior:
Does not let me use an imatrix file.

Model and operation involved:
NA

Diagnostics and logs

GGUF export failed: Unsloth: imatrix_file=True but no upstream Unsloth imatrix was found. Searched repos: (none derived from the base model) Searched files: ['imatrix_unsloth.dat', 'imatrix_unsloth.gguf_file'] Pass imatrix_file='/path/to/imatrix.(dat|gguf)' to use your own.

(But I cannot "use my own" because it appears the software is only looking on HF, not in local folders where I have placed the 'imatrix_unsloth.gguf' file.)

Desktop application

NA

Studio web UI

NA

Model-specific logs, if applicable

NA

Python package or notebook, if applicable

NA

Remove all tokens and private data.

Also include Python, Unsloth, unsloth_zoo, PyTorch, Transformers, and TRL
versions, plus `nvidia-smi` output when applicable.

## Additional context

^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\XXX\\.unsloth\\studio\\unsloth_studio\\Lib\\site-packages\\unsloth\\save.py\", line 4717, in unsloth_save_pretrained_gguf\n    imatrix_path = _resolve_imatrix_file(self, imatrix_file, token, save_directory)\n                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n  File \"C:\\Users\\XXX\\.unsloth\\studio\\unsloth_studio\\Lib\\site-packages\\unsloth\\save.py\", line 5813, in _resolve_imatrix_file\n    raise RuntimeError(\nRuntimeError: Unsloth: imatrix_file=True but no upstream Unsloth imatrix was found.\n  Searched repos: (none derived from the base model)\n  Searched files: ['imatrix_unsloth.dat', 'imatrix_unsloth.gguf_file']\nPass imatrix_file='/path/to/imatrix.(dat|gguf)' to use your own.\n"}
{"timestamp": "2026-08-25T20:07:12.391533Z", "level": "info", "event": "Received command: cleanup"}
{"timestamp": "2026-08-25T20:07:12.391533Z", "level": "info", "event": "Starting memory cleanup..."}