Outmake Models
Models

Phi 3 Medium 4K

Microsoft · May 2024

Get Hugging Face

A 14B chat model from Microsoft.

Rating
2.6
Downloads
8.3K
Likes
39
Parameters
14B
Context
4K
Licence
mit

Runs great on M4 · 16 GB

The Q3_K_M size from bartowski is the one to get: a 6.9 GB download that uses 7.5 GB of memory, and writes about 5 words a second.

Ratings

2.6 out of 5, across people's blind votes on its answers and the tests below.

Text
1198
Coding
1231

LM Arena ratings of October 28, 2026 (CC BY 4.0): people vote between two answers without knowing which model wrote which. The bar ends at 1488, the best any open model has. The Arena tests the full model; a small size on a Mac answers a little worse.

KnowledgeMMLU-Pro
55.7%

MMLU-Pro: graduate-level questions in 14 subjects. Scores are the ones filed on the model's Hugging Face page or the board's own table, read 2026-11-05; some come from the maker.

Phi 3 Medium 4K runs well from a Mac with an M1 and 8 GB of memory. 59 of 59 Mac configurations run it well.

Compare with

The nearest models in size.

Sizes

From bartowski/Phi-3-medium-4k-instruct-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

SizeDownloadMemory at 8KGet
IQ2_M4.7 GB5.8 GBestimatedOpen
Q2_K5.1 GB6.2 GBestimatedOpen
Q2_K_L5.3 GB6.4 GBestimatedOpen
IQ3_XS5.8 GB6.9 GBestimatedOpen
Q3_K_S6.1 GB7.1 GBestimatedOpen
IQ3_M6.5 GB7.6 GBestimatedOpen
Q3_K_M6.9 GB8.0 GBestimatedOpen
IQ4_XS7.5 GB8.5 GBestimatedOpen
Q3_K_L7.5 GB8.6 GBestimatedOpen
Q3_K_XL7.6 GB8.7 GBestimatedOpen
Q4_K_S8.0 GB9.0 GBestimatedOpen
Q4_K_M8.6 GB9.6 GBestimatedOpen
Q4_K_L8.7 GB9.8 GBestimatedOpen
Q5_K_S9.6 GB11 GBestimatedOpen
Q5_K_M10 GB11 GBestimatedOpen
Q5_K_L10 GB11 GBestimatedOpen
Q6_K11 GB13 GBestimatedOpen
Q6_K_L12 GB13 GBestimatedOpen
Q8_015 GB16 GBestimatedOpen
F3256 GB57 GBestimatedOpen

Run It Anywhere

Phi 3 Medium 4K is an open model, so any app that runs GGUF files can run it. These take the Q3_K_M size from bartowski straight from Hugging Face.

  • LM StudioMac, Windows, LinuxOpen
  • JanMac, Windows, LinuxOpen
  • MstyMac, Windows, LinuxOpen
  • OllamaMac, Windows, Linuxollama run hf.co/bartowski/Phi-3-medium-4k-instruct-GGUF:Q3_K_M
  • llama.cppMac, Windows, Linuxllama-server -hf bartowski/Phi-3-medium-4k-instruct-GGUF:Q3_K_M
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/bartowski/Phi-3-medium-4k-instruct-GGUF:Q3_K_M
On a PC Graphics Card
Video memoryLargest size that fitsDownloadMemory at 8K
6 GBNone worth runningneeds 6.9 GB
8 GBIQ3_M6.5 GB7.6 GB
10 GBQ4_K_L8.7 GB9.8 GB
11 GBQ5_K_S9.6 GB11 GB
12 GBQ5_K_L10 GB11 GB
16 GBQ6_K_L12 GB13 GB
20 GBQ8_015 GB16 GB
24 GBQ8_015 GB16 GB
32 GBQ8_015 GB16 GB
48 GBQ8_015 GB16 GB

The largest size, of those that keep three bits a weight or more, whose memory at an 8K chat is under 92% of the card's; the rest is the desktop's. Estimated from the files, the same way as for a Mac; speed on a PC is not estimated yet. A size that doesn't fit the card still runs, slower, with part of it in ordinary memory.

On every Mac
MacSizeFitMemoryWrites
M6 · 16 GBQ3_K_MRuns great7.5 GB7.1
M6 · 24 GBQ5_K_LRuns great11 GB5.7
M6 · 32 GBQ8_0Runs great15 GB5.3
M5 · 16 GBQ3_K_MRuns great7.5 GB6.5
M5 · 24 GBQ5_K_LRuns great11 GB5.2
M5 · 32 GBQ8_0Runs great15 GB4.8
M5 Pro · 24 GBQ5_K_LRuns great11 GB10
M5 Pro · 48 GBQ8_0Runs great15 GB9.6
M5 Pro · 64 GBQ8_0Runs great15 GB9.6
M5 Max · 36 GBQ8_0Runs great15 GB14
M5 Max · 48 GBQ8_0Runs great15 GB19
M5 Max · 64 GBQ8_0Runs great15 GB19
M5 Max · 128 GBQ8_0Runs great15 GB19
M5 Ultra · 96 GBQ8_0Runs great15 GB36
M5 Ultra · 256 GBQ8_0Runs great15 GB36
M5 Ultra · 512 GBQ8_0Runs great15 GB36
M4 · 16 GBQ3_K_MRuns great7.5 GB5
M4 · 24 GBQ5_K_LRuns great11 GB4.1
M4 · 32 GBQ8_0Runs great15 GB3.7
M4 Pro · 24 GBQ5_K_LRuns great11 GB8.9
M4 Pro · 48 GBQ8_0Runs great15 GB8.1
M4 Pro · 64 GBQ8_0Runs great15 GB8.1
M4 Max · 36 GBQ8_0Runs great15 GB12
M4 Max · 48 GBQ8_0Runs great15 GB15
M4 Max · 64 GBQ8_0Runs great15 GB15
M4 Max · 96 GBQ8_0Runs great15 GB12
M4 Max · 128 GBQ8_0Runs great15 GB15
M3 · 8 GBIQ1_MRuns great3.8 GB5.8
M3 · 16 GBQ3_K_MRuns great7.5 GB4.3
M3 · 24 GBQ5_K_LRuns great11 GB3.4
M3 Pro · 18 GBQ4_K_SRuns great8.5 GB5.8
M3 Pro · 36 GBQ8_0Runs great15 GB4.6
M3 Max · 36 GBQ8_0Runs great15 GB8.9
M3 Max · 48 GBQ8_0Runs great15 GB12
M3 Max · 64 GBQ8_0Runs great15 GB12
M3 Max · 96 GBQ8_0Runs great15 GB8.9
M3 Max · 128 GBQ8_0Runs great15 GB12
M3 Ultra · 96 GBQ8_0Runs great15 GB21
M3 Ultra · 256 GBQ8_0Runs great15 GB21
M3 Ultra · 512 GBQ8_0Runs great15 GB21
M2 · 8 GBIQ1_MRuns great3.8 GB5.8
M2 · 16 GBQ3_K_MRuns great7.5 GB4.3
M2 · 24 GBQ5_K_LRuns great11 GB3.4
M2 Pro · 16 GBQ3_K_MRuns great7.5 GB8.2
M2 Pro · 32 GBQ8_0Runs great15 GB6
M2 Max · 32 GBQ8_0Runs great15 GB12
M2 Max · 64 GBQ8_0Runs great15 GB12
M2 Max · 96 GBQ8_0Runs great15 GB12
M2 Ultra · 64 GBQ8_0Runs great15 GB21
M2 Ultra · 128 GBQ8_0Runs great15 GB21
M2 Ultra · 192 GBQ8_0Runs great15 GB21
M1 · 8 GBIQ1_MRuns great3.8 GB3.9
M1 · 16 GBQ3_K_MRuns great7.5 GB2.9
M1 Pro · 16 GBQ3_K_MRuns great7.5 GB7.9
M1 Pro · 32 GBQ8_0Runs great15 GB5.9
M1 Max · 32 GBQ8_0Runs great15 GB11
M1 Max · 64 GBQ8_0Runs great15 GB11
M1 Ultra · 64 GBQ8_0Runs great15 GB19
M1 Ultra · 128 GBQ8_0Runs great15 GB19

Writes is words a second, estimated from the Mac's memory speed and the model's size; how, and how wrong.

Packaged by

2 repositories on Hugging Face carry this model for llama.cpp. Outmake reads them all as one model.

Information

Maker
Microsoft
Released
May 2024
Parameters
14B
Longest chat
4K tokens
Reads pictures
No
Licence
mit
Downloads this month
8,260
Likes
39
Sizes
20, 4.7 to 56 GB
Kind
Maker's release

Badge for the model card

Shows on Phi 3 Medium 4K's Hugging Face card which Macs run it well, linked to this page. It updates every morning.

Runs on Mac badge

Files, downloads and likes read from Hugging Face on November 5, 2026 and refreshed every morning. Memory and speed are estimates until someone running Outmake measures this model on that Mac.

Your Computer

Every model is judged for this Mac. Both are under About This Mac, in the Apple menu.

Every Mac · Every graphics card

Civitai

Open in OutmakeHugging Face

New to Outmake? Download the free beta