Outmake Models
Models

Granite 4.0 H Small

IBM · September 2025

Get Hugging Face

A 32B, 8.8B active chat model from IBM.

Rating
2.9
Downloads
10K
Likes
75
Parameters
32B, 8.8B active
Context
1M
Licence
apache-2.0

Runs well on M4 · 16 GB

The UD-IQ1_M size from unsloth is the one to get: a 10 GB download that uses 10 GB of memory, and writes about 13 words a second.

Ratings

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

Text
1287
Coding
1330

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 4.0 H Small runs well from a Mac with an M1 and 16 GB of memory. 56 of 59 Mac configurations run it well.

Compare with

The nearest models in size.

Sizes

From ibm-granite/granite-4.0-h-small-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

SizeDownloadMemory at 8KGet
Q2_K12 GB12 GBestimatedOpen
Q3_K_S14 GB14 GBestimatedOpen
Q3_K_M15 GB16 GBestimatedOpen
Q3_K_L16 GB17 GBestimatedOpen
Q4_018 GB18 GBestimatedOpen
Q4_K_S18 GB19 GBestimatedOpen
Q4_K_M19 GB20 GBestimatedOpen
Q4_120 GB20 GBestimatedOpen
Q5_022 GB22 GBestimatedOpen
Q5_K_S22 GB22 GBestimatedOpen
Q5_K_M23 GB23 GBestimatedOpen
Q5_124 GB24 GBestimatedOpen
Q6_K26 GB27 GBestimatedOpen
Q8_034 GB34 GBestimatedOpen
F1664 GB65 GBestimatedOpen

Run It Anywhere

Granite 4.0 H Small is an open model, so any app that runs GGUF files can run it. These take the UD-IQ1_M size from unsloth straight from Hugging Face.

  • LM StudioMac, Windows, LinuxOpen
  • JanMac, Windows, LinuxOpen
  • MstyMac, Windows, LinuxOpen
  • OllamaMac, Windows, Linuxollama run hf.co/unsloth/granite-4.0-h-small-GGUF:UD-IQ1_M
  • llama.cppMac, Windows, Linuxllama-server -hf unsloth/granite-4.0-h-small-GGUF:UD-IQ1_M
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/unsloth/granite-4.0-h-small-GGUF:UD-IQ1_M
On a PC Graphics Card
Video memoryLargest size that fitsDownloadMemory at 8K
6 GBNone worth runningneeds 14 GB
8 GBNone worth runningneeds 14 GB
10 GBNone worth runningneeds 14 GB
11 GBNone worth runningneeds 14 GB
12 GBNone worth runningneeds 14 GB
16 GBQ3_K_M15 GB16 GB
20 GBQ4_K_M19 GB20 GB
24 GBQ5_K_M23 GB23 GB
32 GBQ6_K26 GB27 GB
48 GBQ8_034 GB34 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 GBUD-IQ1_MRuns well10 GB19
M6 · 24 GBQ3_K_MRuns well16 GB13
M6 · 32 GBQ4_1Runs well21 GB12
M5 · 16 GBUD-IQ1_MRuns well10 GB18
M5 · 24 GBQ3_K_MRuns well16 GB12
M5 · 32 GBQ4_1Runs well21 GB11
M5 Pro · 24 GBQ3_K_MRuns well16 GB24
M5 Pro · 48 GBQ5_K_MRuns great23 GB17
M5 Pro · 64 GBUD-Q6_K_XLRuns great30 GB14
M5 Max · 36 GBQ3_K_LRuns great17 GB32
M5 Max · 48 GBQ5_K_MRuns great23 GB32
M5 Max · 64 GBUD-Q6_K_XLRuns great30 GB27
M5 Max · 128 GBQ8_0Runs great44 GB27
M5 Ultra · 96 GBQ8_0Runs great44 GB50
M5 Ultra · 256 GBQ8_0Runs great44 GB50
M5 Ultra · 512 GBQ8_0Runs great44 GB50
M4 · 16 GBUD-IQ1_MRuns well10 GB13
M4 · 24 GBQ3_K_MRuns well16 GB9.2
M4 · 32 GBQ4_1Runs well21 GB8.4
M4 Pro · 24 GBQ3_K_MRuns well16 GB18
M4 Pro · 48 GBQ5_K_MRuns great23 GB14
M4 Pro · 64 GBUD-Q6_K_XLRuns great30 GB12
M4 Max · 36 GBQ3_K_LRuns great17 GB23
M4 Max · 48 GBQ5_K_MRuns great23 GB23
M4 Max · 64 GBUD-Q6_K_XLRuns great30 GB20
M4 Max · 96 GBQ8_0Runs great44 GB16
M4 Max · 128 GBQ8_0Runs great44 GB20
M3 · 8 GBUD-TQ1_0Too big19 GB
M3 · 16 GBUD-IQ1_MRuns well10 GB11
M3 · 24 GBQ3_K_MRuns well16 GB8
M3 Pro · 18 GBQ2_KRuns well12 GB14
M3 Pro · 36 GBQ3_K_LRuns great17 GB11
M3 Max · 36 GBQ3_K_LRuns great17 GB19
M3 Max · 48 GBQ5_K_MRuns great23 GB19
M3 Max · 64 GBUD-Q6_K_XLRuns great30 GB17
M3 Max · 96 GBQ8_0Runs great44 GB13
M3 Max · 128 GBQ8_0Runs great44 GB16
M3 Ultra · 96 GBQ8_0Runs great44 GB28
M3 Ultra · 256 GBQ8_0Runs great44 GB28
M3 Ultra · 512 GBQ8_0Runs great44 GB28
M2 · 8 GBUD-TQ1_0Too big19 GB
M2 · 16 GBUD-IQ1_MRuns well10 GB11
M2 · 24 GBQ3_K_MRuns well16 GB8
M2 Pro · 16 GBUD-IQ1_MRuns well10 GB20
M2 Pro · 32 GBQ4_1Runs well21 GB13
M2 Max · 32 GBQ4_1Runs well21 GB24
M2 Max · 64 GBUD-Q6_K_XLRuns great30 GB16
M2 Max · 96 GBQ8_0Runs great44 GB16
M2 Ultra · 64 GBUD-Q6_K_XLRuns great30 GB28
M2 Ultra · 128 GBQ8_0Runs great44 GB28
M2 Ultra · 192 GBQ8_0Runs great44 GB28
M1 · 8 GBUD-TQ1_0Too big19 GB
M1 · 16 GBUD-IQ1_MRuns well10 GB7.4
M1 Pro · 16 GBUD-IQ1_MRuns well10 GB17
M1 Pro · 32 GBQ4_1Runs well21 GB12
M1 Max · 32 GBQ4_1Runs well21 GB20
M1 Max · 64 GBUD-Q6_K_XLRuns great30 GB15
M1 Ultra · 64 GBUD-Q6_K_XLRuns great30 GB23
M1 Ultra · 128 GBQ8_0Runs great44 GB23

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
September 2025
Parameters
32B, 8.8B active
Longest chat
1M tokens
Reads pictures
No
Licence
apache-2.0
Downloads this month
10,373
Likes
75
Sizes
15, 12 to 64 GB
Kind
Maker's release

Badge for the model card

Shows on Granite 4.0 H Small'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