tensorblock/phi-4-GGUF

Feedback and support: TensorBlock's Twitter/X, Telegram Group and Discord server

microsoft/phi-4 - GGUF

This repo contains GGUF format model files for microsoft/phi-4.

The files were quantized using machines provided by TensorBlock, and they are compatible with llama.cpp as of commit b4446.

Run them on the TensorBlock client using your local machine ↗

Prompt template

<|im_start|>system<|im_sep|>{system_prompt}<|im_end|><|im_start|>user<|im_sep|>{prompt}<|im_end|><|im_start|>assistant<|im_sep|>

Model file specification

Filename	Quant type	File Size	Description
phi-4-Q2_K.gguf	Q2_K	5.547 GB	smallest, significant quality loss - not recommended for most purposes
phi-4-Q3_K_S.gguf	Q3_K_S	6.505 GB	very small, high quality loss
phi-4-Q3_K_M.gguf	Q3_K_M	7.363 GB	very small, high quality loss
phi-4-Q3_K_L.gguf	Q3_K_L	7.930 GB	small, substantial quality loss
phi-4-Q4_0.gguf	Q4_0	8.383 GB	legacy; small, very high quality loss - prefer using Q3_K_M
phi-4-Q4_K_S.gguf	Q4_K_S	8.441 GB	small, greater quality loss
phi-4-Q4_K_M.gguf	Q4_K_M	9.053 GB	medium, balanced quality - recommended
phi-4-Q5_0.gguf	Q5_0	10.152 GB	legacy; medium, balanced quality - prefer using Q4_K_M
phi-4-Q5_K_S.gguf	Q5_K_S	10.152 GB	large, low quality loss - recommended
phi-4-Q5_K_M.gguf	Q5_K_M	10.604 GB	large, very low quality loss - recommended
phi-4-Q6_K.gguf	Q6_K	12.030 GB	very large, extremely low quality loss
phi-4-Q8_0.gguf	Q8_0	15.581 GB	very large, extremely low quality loss - not recommended

Downloading instruction

Command line

Firstly, install Huggingface Client

pip install -U "huggingface_hub[cli]"

Then, downoad the individual model file the a local directory

huggingface-cli download tensorblock/phi-4-GGUF --include "phi-4-Q2_K.gguf" --local-dir MY_LOCAL_DIR

If you wanna download multiple model files with a pattern (e.g., *Q4_K*gguf), you can try:

huggingface-cli download tensorblock/phi-4-GGUF --local-dir MY_LOCAL_DIR --local-dir-use-symlinks False --include='*Q4_K*gguf'