metadata

base_model: microsoft/Phi-4-mini-instruct
license: mit
license_link: https://huggingface.co/microsoft/Phi-4-mini-instruct/resolve/main/LICENSE
language:
  - multilingual
  - ar
  - zh
  - cs
  - da
  - nl
  - en
  - fi
  - fr
  - de
  - he
  - hu
  - it
  - ja
  - ko
  - 'no'
  - pl
  - pt
  - ru
  - es
  - sv
  - th
  - tr
  - uk
pipeline_tag: text-generation
library_name: transformers
model_creator: Microsoft
model_name: Phi-4-mini-instruct
quantized_by: Second State Inc.
tags:
  - nlp
  - code

Phi-4-mini-instruct-GGUF

Original Model

microsoft/Phi-4-mini-instruct

Run with LlamaEdge

LlamaEdge version: coming soon

Prompt template

Prompt type: phi-4-chat

Prompt string

<|system|>Insert System Message<|end|><|user|>Insert User Message<|end|><|assistant|>

Context size: 128000

Run as LlamaEdge service

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Phi-4-mini-instruct-Q5_K_M.gguf \
  llama-api-server.wasm \
  --prompt-template phi-4-chat \
  --ctx-size 128000 \
  --model-name phi-4-mini

Run as LlamaEdge command app

wasmedge --dir .:. --nn-preload default:GGML:AUTO:Phi-4-mini-instruct-Q5_K_M.gguf \
  llama-chat.wasm \
  --prompt-template phi-4-chat \
  --ctx-size 128000

Quantized GGUF Models

Name	Quant method	Bits	Size	Use case
Phi-4-mini-instruct-Q2_K.gguf	Q2_K	2	1.68 GB	smallest, significant quality loss - not recommended for most purposes
Phi-4-mini-instruct-Q3_K_L.gguf	Q3_K_L	3	2.25 GB	small, substantial quality loss
Phi-4-mini-instruct-Q3_K_M.gguf	Q3_K_M	3	2.12 GB	very small, high quality loss
Phi-4-mini-instruct-Q3_K_S.gguf	Q3_K_S	3	1.90 GB	very small, high quality loss
Phi-4-mini-instruct-Q4_0.gguf	Q4_0	4	2.33 GB	legacy; small, very high quality loss - prefer using Q3_K_M
Phi-4-mini-instruct-Q4_K_M.gguf	Q4_K_M	4	2.49 GB	medium, balanced quality - recommended
Phi-4-mini-instruct-Q4_K_S.gguf	Q4_K_S	4	2.34 GB	small, greater quality loss
Phi-4-mini-instruct-Q5_0.gguf	Q5_0	5	2.73 GB	legacy; medium, balanced quality - prefer using Q4_K_M
Phi-4-mini-instruct-Q5_K_M.gguf	Q5_K_M	5	2.85 GB	large, very low quality loss - recommended
Phi-4-mini-instruct-Q5_K_S.gguf	Q5_K_S	5	2.73 GB	large, low quality loss - recommended
Phi-4-mini-instruct-Q6_K.gguf	Q6_K	6	3.16 GB	very large, extremely low quality loss
Phi-4-mini-instruct-Q8_0.gguf	Q8_0	8	4.08 GB	very large, extremely low quality loss - not recommended
Phi-4-mini-instruct-f16.gguf	f16	16	7.68 GB

Quantized with llama.cpp b4792.