Outmake Models
Models

Gemma 3 27B

Google · March 2025

Get Hugging Face

A 27B chat model from Google that reads pictures as well as text.

Rating
2.8
Downloads
115K
Likes
447
Parameters
27B
Context
128K
Licence
gemma

Runs great on M4 · 16 GB

The UD-IQ1_M size from unsloth is the one to get: a 7.0 GB download that uses 7.7 GB of memory, and writes about 6.7 words a second.

Ratings

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

Text
1365
Coding
1357
Vision
1161

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.

CodingAider polyglot
4.9%

Aider polyglot: coding exercises solved in six languages. 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.

Gemma 3 27B 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 unsloth/gemma-3-27b-it-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

SizeDownloadMemory at 8KGet
UD-IQ1_S6.5 GB7.2 GBestimatedOpen
UD-IQ1_M7.0 GB7.7 GBestimatedOpen
UD-IQ2_XXS7.9 GB8.6 GBestimatedOpen
UD-IQ2_M9.6 GB10 GBestimatedOpen
Q2_K11 GB11 GBestimatedOpen
Q2_K_L11 GB11 GBestimatedOpen
UD-Q2_K_XL11 GB11 GBestimatedOpen
UD-IQ3_XXS11 GB12 GBestimatedOpen
Q3_K_S12 GB13 GBestimatedOpen
Q3_K_M13 GB14 GBestimatedOpen
UD-Q3_K_XL14 GB14 GBestimatedOpen
IQ4_XS15 GB15 GBestimatedOpen
IQ4_NL16 GB16 GBestimatedOpen
Q4_016 GB16 GBestimatedOpen
Q4_K_S16 GB16 GBestimatedOpen
Q4_K_M17 GB17 GBestimatedOpen
UD-Q4_K_XL17 GB18 GBestimatedOpen
Q4_117 GB18 GBestimatedOpen
Q5_K_S19 GB19 GBestimatedOpen
Q5_K_M19 GB20 GBestimatedOpen
UD-Q5_K_XL19 GB20 GBestimatedOpen
Q6_K22 GB23 GBestimatedOpen
UD-Q6_K_XL24 GB24 GBestimatedOpen
Q8_029 GB29 GBestimatedOpen
UD-Q8_K_XL32 GB33 GBestimatedOpen
BF1654 GB55 GBestimatedOpen

Run It Anywhere

Gemma 3 27B 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/gemma-3-27b-it-GGUF:UD-IQ1_M
  • llama.cppMac, Windows, Linuxllama-server -hf unsloth/gemma-3-27b-it-GGUF:UD-IQ1_M
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/unsloth/gemma-3-27b-it-GGUF:UD-IQ1_M
On a PC Graphics Card
Video memoryLargest size that fitsDownloadMemory at 8K
6 GBNone worth runningneeds 10 GB
8 GBNone worth runningneeds 10 GB
10 GBNone worth runningneeds 10 GB
11 GBUD-IQ2_M9.6 GB10 GB
12 GBUD-IQ3_XXS11 GB12 GB
16 GBIQ4_XS15 GB15 GB
20 GBQ5_K_S19 GB19 GB
24 GBQ6_K22 GB23 GB
32 GBQ8_029 GB29 GB
48 GBUD-Q8_K_XL32 GB33 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 great7.7 GB9.6
M6 · 24 GBQ3_K_XLRuns well16 GB5.4
M6 · 32 GBQ3_K_LRuns great15 GB5.5
M5 · 16 GBUD-IQ1_MRuns great7.7 GB8.8
M5 · 24 GBQ3_K_XLRuns well16 GB4.9
M5 · 32 GBQ3_K_LRuns great15 GB5
M5 Pro · 24 GBQ3_K_XLRuns well16 GB9.6
M5 Pro · 48 GBQ6_KRuns great23 GB7.2
M5 Pro · 64 GBQ8_0Runs great31 GB6.5
M5 Max · 36 GBQ4_K_MRuns great17 GB13
M5 Max · 48 GBQ6_KRuns great23 GB14
M5 Max · 64 GBQ8_0Runs great31 GB13
M5 Max · 128 GBQ8_0Runs great35 GB13
M5 Ultra · 96 GBQ8_0Runs great35 GB25
M5 Ultra · 256 GBQ8_0Runs great35 GB25
M5 Ultra · 512 GBQ8_0Runs great35 GB25
M4 · 16 GBUD-IQ1_MRuns great7.7 GB6.7
M4 · 24 GBQ3_K_XLRuns well16 GB3.8
M4 · 32 GBQ3_K_LRuns great15 GB3.9
M4 Pro · 24 GBQ3_K_XLRuns well16 GB8.1
M4 Pro · 48 GBQ6_KRuns great23 GB6
M4 Pro · 64 GBQ8_0Runs great31 GB5.5
M4 Max · 36 GBQ4_K_MRuns great17 GB11
M4 Max · 48 GBQ6_KRuns great23 GB11
M4 Max · 64 GBQ8_0Runs great31 GB10
M4 Max · 96 GBQ8_0Runs great35 GB7.9
M4 Max · 128 GBQ8_0Runs great35 GB10
M3 · 8 GBUD-IQ1_SToo big13 GB
M3 · 16 GBUD-IQ1_MRuns great7.7 GB5.7
M3 · 24 GBQ3_K_XLRuns well16 GB3.3
M3 Pro · 18 GBUD-IQ3_XXSRuns well12 GB6.1
M3 Pro · 36 GBQ4_K_MRuns great17 GB4.4
M3 Max · 36 GBQ4_K_MRuns great17 GB8.3
M3 Max · 48 GBQ6_KRuns great23 GB8.6
M3 Max · 64 GBQ8_0Runs great31 GB7.9
M3 Max · 96 GBQ8_0Runs great35 GB6
M3 Max · 128 GBQ8_0Runs great35 GB7.9
M3 Ultra · 96 GBQ8_0Runs great35 GB15
M3 Ultra · 256 GBQ8_0Runs great35 GB15
M3 Ultra · 512 GBQ8_0Runs great35 GB15
M2 · 8 GBUD-IQ1_SToo big13 GB
M2 · 16 GBUD-IQ1_MRuns great7.7 GB5.7
M2 · 24 GBQ3_K_XLRuns well16 GB3.2
M2 Pro · 16 GBUD-IQ1_MRuns great7.7 GB11
M2 Pro · 32 GBQ3_K_LRuns great15 GB6.4
M2 Max · 32 GBQ3_K_LRuns great15 GB12
M2 Max · 64 GBQ8_0Runs great31 GB7.9
M2 Max · 96 GBQ8_0Runs great35 GB7.9
M2 Ultra · 64 GBQ8_0Runs great31 GB14
M2 Ultra · 128 GBQ8_0Runs great35 GB14
M2 Ultra · 192 GBQ8_0Runs great35 GB14
M1 · 8 GBUD-IQ1_SToo big13 GB
M1 · 16 GBUD-IQ1_MRuns great7.7 GB3.9
M1 Pro · 16 GBUD-IQ1_MRuns great7.7 GB9.8
M1 Pro · 32 GBQ3_K_LRuns great15 GB6
M1 Max · 32 GBQ3_K_LRuns great15 GB11
M1 Max · 64 GBQ8_0Runs great31 GB7.5
M1 Ultra · 64 GBQ8_0Runs great31 GB13
M1 Ultra · 128 GBQ8_0Runs great35 GB13

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

Packaged by

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

Information

Maker
Google
Released
March 2025
Parameters
27B
Longest chat
128K tokens
Reads pictures
Yes
Licence
gemma
Downloads this month
114,615
Likes
447
Sizes
26, 6.5 to 54 GB
Kind
Maker's release

Badge for the model card

Shows on Gemma 3 27B'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