Outmake Models
Models

G4 LuneLight 31B

zerofata · October 2026

Get Hugging Face

A 31B chat model from zerofata.

Downloads
17K
Likes
22
Parameters
31B
Context
256K
Licence
apache-2.0

Runs well on M4 · 16 GB

The IQ2_XS size from mradermacher is the one to get: a 9.5 GB download that uses 10 GB of memory, and writes about 5.3 words a second.

G4 LuneLight 31B 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 zerofata/G4-LuneLight-31B-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

SizeDownloadMemory at 8KGet
IQ3_M14 GB15 GBestimatedOpen
Q3_K_M15 GB16 GBestimatedOpen
IQ4_XS17 GB18 GBestimatedOpen
Q4_K_M19 GB20 GBestimatedOpen
Q5_K_M22 GB23 GBestimatedOpen
Q6_K25 GB26 GBestimatedOpen
Q8_033 GB34 GBestimatedOpen
BF1661 GB62 GBestimatedOpen

Run It Anywhere

G4 LuneLight 31B is an open model, so any app that runs GGUF files can run it. These take the IQ2_XS size from mradermacher straight from Hugging Face.

  • LM StudioMac, Windows, LinuxOpen
  • JanMac, Windows, LinuxOpen
  • MstyMac, Windows, LinuxOpen
  • OllamaMac, Windows, Linuxollama run hf.co/mradermacher/G4-LuneLight-31B-i1-GGUF:IQ2_XS
  • llama.cppMac, Windows, Linuxllama-server -hf mradermacher/G4-LuneLight-31B-i1-GGUF:IQ2_XS
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/mradermacher/G4-LuneLight-31B-i1-GGUF:IQ2_XS
On a PC Graphics Card
Video memoryLargest size that fitsDownloadMemory at 8K
6 GBNone worth runningneeds 15 GB
8 GBNone worth runningneeds 15 GB
10 GBNone worth runningneeds 15 GB
11 GBNone worth runningneeds 15 GB
12 GBNone worth runningneeds 15 GB
16 GBIQ3_M14 GB15 GB
20 GBQ4_K_M19 GB20 GB
24 GBQ5_K_M22 GB23 GB
32 GBQ6_K25 GB26 GB
48 GBQ8_033 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 GBIQ2_XSRuns well10 GB7.5
M6 · 24 GBIQ3_MRuns well16 GB5.4
M6 · 32 GBQ4_K_MRuns well21 GB4.4
M5 · 16 GBIQ2_XSRuns well10 GB6.9
M5 · 24 GBIQ3_MRuns well16 GB5
M5 · 32 GBQ4_K_MRuns well21 GB4
M5 Pro · 24 GBIQ3_MRuns well16 GB9.8
M5 Pro · 48 GBQ5_K_MRuns great23 GB6.9
M5 Pro · 64 GBQ6_KRuns great29 GB6.4
M5 Max · 36 GBQ3_K_MRuns great17 GB14
M5 Max · 48 GBQ5_K_MRuns great23 GB13
M5 Max · 64 GBQ6_KRuns great29 GB13
M5 Max · 128 GBQ8_0Runs great47 GB11
M5 Ultra · 96 GBQ8_0Runs great40 GB22
M5 Ultra · 256 GBQ8_0Runs great47 GB22
M5 Ultra · 512 GBQ8_0Runs great47 GB22
M4 · 16 GBIQ2_XSRuns well10 GB5.3
M4 · 24 GBIQ3_MRuns well16 GB3.9
M4 · 32 GBQ4_K_MRuns well21 GB3.1
M4 Pro · 24 GBIQ3_MRuns well16 GB8.3
M4 Pro · 48 GBQ5_K_MRuns great23 GB6
M4 Pro · 64 GBQ6_KRuns great29 GB5.3
M4 Max · 36 GBQ3_K_MRuns great17 GB11
M4 Max · 48 GBQ5_K_MRuns great23 GB11
M4 Max · 64 GBQ6_KRuns great29 GB9.9
M4 Max · 96 GBQ8_0Runs great40 GB7.1
M4 Max · 128 GBQ8_0Runs great47 GB9.2
M3 · 8 GBIQ1_SToo big21 GB
M3 · 16 GBIQ2_XSRuns well10 GB4.5
M3 · 24 GBIQ3_MRuns well16 GB3.3
M3 Pro · 18 GBIQ1_MRuns great8.7 GB7.6
M3 Pro · 36 GBQ3_K_MRuns great17 GB4.6
M3 Max · 36 GBQ3_K_MRuns great17 GB8.8
M3 Max · 48 GBQ5_K_MRuns great23 GB8.6
M3 Max · 64 GBQ6_KRuns great29 GB7.7
M3 Max · 96 GBQ8_0Runs great40 GB5.4
M3 Max · 128 GBQ8_0Runs great47 GB7
M3 Ultra · 96 GBQ8_0Runs great40 GB13
M3 Ultra · 256 GBQ8_0Runs great47 GB13
M3 Ultra · 512 GBQ8_0Runs great47 GB13
M2 · 8 GBIQ1_SToo big21 GB
M2 · 16 GBIQ2_XSRuns well10 GB4.5
M2 · 24 GBIQ3_MRuns well16 GB3.3
M2 Pro · 16 GBIQ2_XSRuns well10 GB8.6
M2 Pro · 32 GBQ4_K_MRuns well21 GB5.1
M2 Max · 32 GBQ4_K_MRuns well21 GB9.7
M2 Max · 64 GBQ6_KRuns great29 GB7.7
M2 Max · 96 GBQ8_0Runs great40 GB7
M2 Ultra · 64 GBQ6_KRuns great29 GB14
M2 Ultra · 128 GBQ8_0Runs great47 GB13
M2 Ultra · 192 GBQ8_0Runs great47 GB13
M1 · 8 GBIQ1_SToo big21 GB
M1 · 16 GBIQ2_XSRuns well10 GB3
M1 Pro · 16 GBIQ2_XSRuns well10 GB8
M1 Pro · 32 GBQ4_K_MRuns well21 GB4.9
M1 Max · 32 GBQ4_K_MRuns well21 GB9
M1 Max · 64 GBQ6_KRuns great29 GB7.2
M1 Ultra · 64 GBQ6_KRuns great29 GB13
M1 Ultra · 128 GBQ8_0Runs great47 GB12

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
zerofata
Released
October 2026
Parameters
31B
Longest chat
256K tokens
Reads pictures
No
Licence
apache-2.0
Downloads this month
17,223
Likes
22
Sizes
8, 14 to 61 GB
Kind
Maker's release

Badge for the model card

Shows on G4 LuneLight 31B'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