Outmake Models
Models

Granite 3.1 2B

IBM · December 2024

Get Hugging Face

A 2.5B chat model from IBM.

Rating
2.2
Downloads
6.1K
Likes
8
Parameters
2.5B
Context
128K
Licence
apache-2.0

Runs great on M4 · 16 GB

The Q8_0 size from bartowski is the one to get: a 2.7 GB download that uses 5.9 GB of memory, and writes about 14 words a second.

Ratings

2.2 out of 5, across people's blind votes on its answers.

Text
1179
Coding
1249

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.

Granite 3.1 2B 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/granite-3.1-2b-instruct-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

SizeDownloadMemory at 8KGet
IQ2_M0.9 GB1.4 GBestimatedOpen
Q2_K1.0 GB1.5 GBestimatedOpen
Q2_K_L1.0 GB1.5 GBestimatedOpen
IQ3_XS1.1 GB1.6 GBestimatedOpen
Q3_K_S1.1 GB1.6 GBestimatedOpen
IQ3_M1.2 GB1.7 GBestimatedOpen
Q3_K_M1.3 GB1.7 GBestimatedOpen
Q3_K_L1.4 GB1.9 GBestimatedOpen
Q3_K_XL1.4 GB1.9 GBestimatedOpen
IQ4_XS1.4 GB1.9 GBestimatedOpen
IQ4_NL1.5 GB2.0 GBestimatedOpen
Q4_01.5 GB2.0 GBestimatedOpen
Q4_K_S1.5 GB2.0 GBestimatedOpen
Q4_K_M1.5 GB2.0 GBestimatedOpen
Q4_K_L1.6 GB2.1 GBestimatedOpen
Q5_K_S1.8 GB2.3 GBestimatedOpen
Q5_K_M1.8 GB2.3 GBestimatedOpen
Q5_K_L1.8 GB2.3 GBestimatedOpen
Q6_K2.1 GB2.6 GBestimatedOpen
Q6_K_L2.1 GB2.6 GBestimatedOpen
Q8_02.7 GB3.2 GBestimatedOpen
F165.1 GB5.6 GBestimatedOpen

Run It Anywhere

Granite 3.1 2B is an open model, so any app that runs GGUF files can run it. These take the Q8_0 size from bartowski straight from Hugging Face.

  • LM StudioMac, Windows, LinuxOpen
  • JanMac, Windows, LinuxOpen
  • MstyMac, Windows, LinuxOpen
  • OllamaMac, Windows, Linuxollama run hf.co/bartowski/granite-3.1-2b-instruct-GGUF:Q8_0
  • llama.cppMac, Windows, Linuxllama-server -hf bartowski/granite-3.1-2b-instruct-GGUF:Q8_0
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/bartowski/granite-3.1-2b-instruct-GGUF:Q8_0
On a PC Graphics Card
Video memoryLargest size that fitsDownloadMemory at 8K
6 GBF165.1 GB5.6 GB
8 GBF165.1 GB5.6 GB
10 GBF165.1 GB5.6 GB
11 GBF165.1 GB5.6 GB
12 GBF165.1 GB5.6 GB
16 GBF165.1 GB5.6 GB
20 GBF165.1 GB5.6 GB
24 GBF165.1 GB5.6 GB
32 GBF165.1 GB5.6 GB
48 GBF165.1 GB5.6 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 GBQ8_0Runs great5.9 GB20
M6 · 24 GBQ8_0Runs great9.0 GB20
M6 · 32 GBQ8_0Runs great9.0 GB20
M5 · 16 GBQ8_0Runs great5.9 GB19
M5 · 24 GBQ8_0Runs great9.0 GB19
M5 · 32 GBQ8_0Runs great9.0 GB19
M5 Pro · 24 GBQ8_0Runs great9.0 GB36
M5 Pro · 48 GBQ8_0Runs great9.0 GB36
M5 Pro · 64 GBQ8_0Runs great9.0 GB36
M5 Max · 36 GBQ8_0Runs great9.0 GB52
M5 Max · 48 GBQ8_0Runs great9.0 GB67
M5 Max · 64 GBQ8_0Runs great9.0 GB67
M5 Max · 128 GBQ8_0Runs great9.0 GB67
M5 Ultra · 96 GBQ8_0Runs great9.0 GB118
M5 Ultra · 256 GBQ8_0Runs great9.0 GB118
M5 Ultra · 512 GBQ8_0Runs great9.0 GB118
M4 · 16 GBQ8_0Runs great5.9 GB14
M4 · 24 GBQ8_0Runs great9.0 GB14
M4 · 32 GBQ8_0Runs great9.0 GB14
M4 Pro · 24 GBQ8_0Runs great9.0 GB27
M4 Pro · 48 GBQ8_0Runs great9.0 GB27
M4 Pro · 64 GBQ8_0Runs great9.0 GB27
M4 Max · 36 GBQ8_0Runs great9.0 GB36
M4 Max · 48 GBQ8_0Runs great9.0 GB44
M4 Max · 64 GBQ8_0Runs great9.0 GB44
M4 Max · 96 GBQ8_0Runs great9.0 GB36
M4 Max · 128 GBQ8_0Runs great9.0 GB44
M3 · 8 GBQ8_0Runs great3.6 GB12
M3 · 16 GBQ8_0Runs great5.9 GB12
M3 · 24 GBQ8_0Runs great9.0 GB12
M3 Pro · 18 GBQ8_0Runs great5.9 GB17
M3 Pro · 36 GBQ8_0Runs great9.0 GB17
M3 Max · 36 GBQ8_0Runs great9.0 GB30
M3 Max · 48 GBQ8_0Runs great9.0 GB37
M3 Max · 64 GBQ8_0Runs great9.0 GB37
M3 Max · 96 GBQ8_0Runs great9.0 GB30
M3 Max · 128 GBQ8_0Runs great9.0 GB37
M3 Ultra · 96 GBQ8_0Runs great9.0 GB59
M3 Ultra · 256 GBQ8_0Runs great9.0 GB59
M3 Ultra · 512 GBQ8_0Runs great9.0 GB59
M2 · 8 GBQ8_0Runs great3.6 GB12
M2 · 16 GBQ8_0Runs great5.9 GB12
M2 · 24 GBQ8_0Runs great9.0 GB12
M2 Pro · 16 GBQ8_0Runs great5.9 GB22
M2 Pro · 32 GBQ8_0Runs great9.0 GB22
M2 Max · 32 GBQ8_0Runs great9.0 GB37
M2 Max · 64 GBQ8_0Runs great9.0 GB37
M2 Max · 96 GBQ8_0Runs great9.0 GB37
M2 Ultra · 64 GBQ8_0Runs great9.0 GB59
M2 Ultra · 128 GBQ8_0Runs great9.0 GB59
M2 Ultra · 192 GBQ8_0Runs great9.0 GB59
M1 · 8 GBQ8_0Runs great3.6 GB7.8
M1 · 16 GBQ8_0Runs great5.9 GB7.8
M1 Pro · 16 GBQ8_0Runs great5.9 GB20
M1 Pro · 32 GBQ8_0Runs great9.0 GB20
M1 Max · 32 GBQ8_0Runs great9.0 GB32
M1 Max · 64 GBQ8_0Runs great9.0 GB32
M1 Ultra · 64 GBQ8_0Runs great9.0 GB46
M1 Ultra · 128 GBQ8_0Runs great9.0 GB46

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
IBM
Released
December 2024
Parameters
2.5B
Longest chat
128K tokens
Reads pictures
No
Licence
apache-2.0
Downloads this month
6,147
Likes
8
Sizes
22, 0.9 to 5.1 GB
Kind
Maker's release

Badge for the model card

Shows on Granite 3.1 2B'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