Downloads · 30 days
53
6% of all-time downloads
phi-4-imatrix-GGUF
liodon-ai/phi-4-imatrix-GGUF
phi-4-imatrix-GGUF is a text generation model from liodon-ai. Use it when you need the model to write or continue text. It is set up for gguf. The card lists the license as other.
GGUF quantizations of microsoft/phi-4, published by Liodon AI.
All-time downloads
932
Public
Repo size
67.8 GB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.gguf67.8 GB · 100%
At a glance
- Task
- Text Generation
- Library
- gguf
- License
- other
- Access
- Public
- Created
- Jul 9, 2026
- Updated
- Jul 13, 2026
- SHA
- a29ba8e2
Try a prompt
Base models
Model card
From the Hugging Face model README
phi-4 — iMatrix GGUF
GGUF quantizations of microsoft/phi-4, published by Liodon AI.
Quick Start
llama.cpp
llama-cli -hf liodon-ai/phi-4-imatrix-GGUF:Q4_K_M
Ollama
ollama run hf.co/liodon-ai/phi-4-imatrix-GGUF:Q4_K_M
LM Studio / Jan — search liodon-ai/phi-4-imatrix-GGUF and pick your quant.
Quants
| Quant | Size | VRAM est. | Notes |
|---|---|---|---|
IQ2_M | 5.49 GB | ~6 GB | 2-bit, iMatrix — smallest usable |
IQ3_M | 6.91 GB | ~8 GB | 3-bit, iMatrix — great quality/size tradeoff |
IQ4_XS | 8.19 GB | ~9 GB | 4-bit extra-small, iMatrix |
Q4_K_M | 9.06 GB | ~10 GB | 4-bit, iMatrix-calibrated (recommended) |
Q5_K_M | 10.50 GB | ~12 GB | 5-bit, iMatrix-calibrated |
Q6_K | 12.03 GB | ~14 GB | 6-bit, iMatrix-calibrated, near-lossless |
Q8_0 | 15.58 GB | ~18 GB | 8-bit, essentially lossless |
What is iMatrix?
Standard quantization treats all weights equally. iMatrix runs 128 calibration chunks through the full-precision model to find which weights matter most, then allocates more precision where it counts. At Q2/Q3/Q4 this means noticeably better coherence and instruction-following — same file size, better output.
Calibration: 2M tokens of WikiText-103.
Also see plain (non-iMatrix) quants:
liodon-ai/phi-4-GGUF
Source
- Model: microsoft/phi-4
- License: other
Quantized by Liodon AI
- Task
- Text Generation
- Library
- gguf
- License
- other
- Created
- Jul 9, 2026
- Updated
- Jul 13, 2026