Phi-4 mini instruct

Microsoft

Free

Get

MLX Score1.5

PriceFreeno paid plan

RunsOn your Mac

DeveloperMicrosoft

LicenceOpenMIT

CategoryModels

Mac14.0+macOS or later

Phi-4 mini instruct is a 3.8-billion-parameter model from Microsoft, trained on high-quality, reasoning-dense data and strong at maths and logic for its size. It has a 128K-token context. Microsoft notes that a model this small can't hold many facts, and suggests pairing it with search.

A 4-bit MLX build is about 2 GB; on mlx.app's estimate it fits a Mac with 8 GB of memory.

Open weights under MIT. Run locally, nothing leaves your Mac.

How to Get It

Run it on your MacSee how

WebsiteVisit

Official downloads only. MLX never re-hosts an app, and never runs a command for you.

MLX Score

1.5out of 5

Hugging Face downloads and likesahead of 13%

Hugging Face: 347,033 downloads (30 days), 878 likes · read 2026-11-02

Ranked against the open models listings, from real public data only. How the score works.

Information

Developer: Microsoft
Category: Open models
Price: Free (Open weights, free to download.)
Your data: Runs on your Mac
Licence: MIT
Mac: macOS 14.0 or later
iPhone and iPad: —
Needs: 8 GB memory
Links: Website · Hugging Face
Facts checked: 2026-11-01

Sources

huggingface.co/microsoft/Phi-4-mini-instruct · MIT, 3.8B parameters, 128K context, reasoning focus, limited factual knowledge · checked 2026-11-01huggingface.co/mlx-community/Phi-4-mini-instruct-4bit · 4-bit MLX build (about 2 GB) · checked 2026-11-01ml-explore.github.io/mlx/build/html/install.html · macOS 14.0 minimum (MLX needs Apple silicon and macOS 14.0+) · checked 2026-11-01mlx.app/run/phi-4-mini · memory: smallest Mac (8 GB) where the 4-bit MLX build fits with room for context, from lib/compare (75% of unified memory usable) · checked 2026-11-01Icon from Hugging Face

See All