Outmake Models
Models

Clef

Cloudflare · September 2026

Get Hugging Face

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

Downloads
61K
Likes
68
Parameters
27B
Context
256K
Licence
apache-2.0

Runs well on M4 · 16 GB

The IQ2_M size from abenzerps is the one to get: a 10 GB download that uses 10 GB of memory, and writes about 5.4 words a second.

Clef 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 bartowski/Cloudflare_clef-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
IQ3_M14 GB14 GBestimatedOpen
Q3_K_M14 GB15 GBestimatedOpen
Q3_K_L15 GB16 GBestimatedOpen
IQ4_XS16 GB16 GBestimatedOpen
Q4_016 GB17 GBestimatedOpen
IQ4_NL16 GB17 GBestimatedOpen
Q4_K_S17 GB17 GBestimatedOpen
Q4_K_M18 GB18 GBestimatedOpen
Q4_118 GB18 GBestimatedOpen
Q4_K_L19 GB19 GBestimatedOpen
Q5_K_S20 GB20 GBestimatedOpen
Q5_K_M21 GB21 GBestimatedOpen
Q6_K23 GB24 GBestimatedOpen
Q6_K_L24 GB24 GBestimatedOpen
Q8_029 GB29 GBestimatedOpen
BF1654 GB54 GBestimatedOpen

Run It Anywhere

Clef is an open model, so any app that runs GGUF files can run it. These take the IQ2_M size from abenzerps straight from Hugging Face.

  • LM StudioMac, Windows, LinuxOpen
  • JanMac, Windows, LinuxOpen
  • MstyMac, Windows, LinuxOpen
  • OllamaMac, Windows, Linuxollama run hf.co/abenzerps/Clef-GGUF:IQ2_M
  • llama.cppMac, Windows, Linuxllama-server -hf abenzerps/Clef-GGUF:IQ2_M
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/abenzerps/Clef-GGUF:IQ2_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_L15 GB16 GB
20 GBQ4_K_L19 GB19 GB
24 GBQ5_K_M21 GB21 GB
32 GBQ8_029 GB29 GB
48 GBQ8_029 GB29 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_MRuns well10 GB7.6
M6 · 24 GBQ3_K_LRuns well16 GB5.4
M6 · 32 GBQ3_K_MRuns great15 GB5.6
M5 · 16 GBIQ2_MRuns well10 GB7
M5 · 24 GBQ3_K_LRuns well16 GB4.9
M5 · 32 GBQ3_K_MRuns great15 GB5.1
M5 Pro · 24 GBQ3_K_LRuns well16 GB9.7
M5 Pro · 48 GBQ5_K_MRuns great22 GB7.4
M5 Pro · 64 GBQ8_0Runs great30 GB6.6
M5 Max · 36 GBQ4_K_SRuns great17 GB13
M5 Max · 48 GBQ5_K_MRuns great22 GB14
M5 Max · 64 GBQ8_0Runs great30 GB13
M5 Max · 128 GBQ8_0Runs great39 GB13
M5 Ultra · 96 GBQ8_0Runs great39 GB25
M5 Ultra · 256 GBQ8_0Runs great39 GB25
M5 Ultra · 512 GBQ8_0Runs great39 GB25
M4 · 16 GBIQ2_MRuns well10 GB5.4
M4 · 24 GBQ3_K_LRuns well16 GB3.9
M4 · 32 GBQ3_K_MRuns great15 GB4
M4 Pro · 24 GBQ3_K_LRuns well16 GB8.2
M4 Pro · 48 GBQ5_K_MRuns great22 GB6.4
M4 Pro · 64 GBQ8_0Runs great30 GB5.5
M4 Max · 36 GBQ4_K_SRuns great17 GB11
M4 Max · 48 GBQ5_K_MRuns great22 GB12
M4 Max · 64 GBQ8_0Runs great30 GB10
M4 Max · 96 GBQ8_0Runs great39 GB8
M4 Max · 128 GBQ8_0Runs great39 GB10
M3 · 8 GBIQ1_MToo big18 GB
M3 · 16 GBIQ2_MRuns well10 GB4.6
M3 · 24 GBQ3_K_LRuns well16 GB3.3
M3 Pro · 18 GBQ2_KRuns well11 GB6.3
M3 Pro · 36 GBQ4_K_SRuns great17 GB4.5
M3 Max · 36 GBQ4_K_SRuns great17 GB8.4
M3 Max · 48 GBQ5_K_MRuns great22 GB9.2
M3 Max · 64 GBQ8_0Runs great30 GB7.9
M3 Max · 96 GBQ8_0Runs great39 GB6.1
M3 Max · 128 GBQ8_0Runs great39 GB7.9
M3 Ultra · 96 GBQ8_0Runs great39 GB15
M3 Ultra · 256 GBQ8_0Runs great39 GB15
M3 Ultra · 512 GBQ8_0Runs great39 GB15
M2 · 8 GBIQ1_MToo big18 GB
M2 · 16 GBIQ2_MRuns well10 GB4.6
M2 · 24 GBQ3_K_LRuns well16 GB3.3
M2 Pro · 16 GBIQ2_MRuns well10 GB8.6
M2 Pro · 32 GBQ3_K_MRuns great15 GB6.5
M2 Max · 32 GBQ3_K_MRuns great15 GB12
M2 Max · 64 GBQ8_0Runs great30 GB7.9
M2 Max · 96 GBQ8_0Runs great39 GB7.9
M2 Ultra · 64 GBQ8_0Runs great30 GB15
M2 Ultra · 128 GBQ8_0Runs great39 GB15
M2 Ultra · 192 GBQ8_0Runs great39 GB15
M1 · 8 GBIQ1_MToo big18 GB
M1 · 16 GBIQ2_MRuns well10 GB3.1
M1 Pro · 16 GBIQ2_MRuns well10 GB8
M1 Pro · 32 GBQ3_K_MRuns great15 GB6.2
M1 Max · 32 GBQ3_K_MRuns great15 GB11
M1 Max · 64 GBQ8_0Runs great30 GB7.5
M1 Ultra · 64 GBQ8_0Runs great30 GB13
M1 Ultra · 128 GBQ8_0Runs great39 GB13

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

Packaged by

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

Information

Maker
Cloudflare
Released
September 2026
Parameters
27B
Longest chat
256K tokens
Reads pictures
Yes
Licence
apache-2.0
Downloads this month
60,742
Likes
68
Sizes
18, 12 to 54 GB
Kind
Maker's release

Badge for the model card

Shows on Clef'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