Skip to content

Phi-4-mini-instruct

A lightweight open model with 128K context

Visit official site

What Phi-4-mini-instruct is

The model is described as a lightweight open-source language model from Microsoft, trained on synthetic and filtered public web data for reasoning-heavy tasks and tuned for instruction following and safety.

Getting started

  • Download the model files from Hugging Face
  • Load it with a framework such as PyTorch and set up inference
  • Choose a chat or tool-calling input format

Pros & cons

✓ What we like

  • Safety evaluation and red teaming are described
  • English, Chinese, Japanese and other languages are listed

! What to watch out for

  • The model is aimed at resource-limited environments
  • Input format is chosen as chat or tool calling

FAQ

How long is the context?

The material states support for up to 128K tokens.

When was it released?

The material names February 2025 as the release date.

Last reviewed: 2026-11-15

More AI coding tools tools

View all →

How we review