Outmake Models
Models

Qwen3 8B

Alibaba Qwen · April 2025

Get Hugging Face

A 8.2B chat model from Alibaba Qwen.

Downloads
1.1M
Likes
617
Parameters
8.2B
Context
40K
Licence
apache-2.0

Runs great on M4 · 16 GB

The Q6_K size from Qwen is the one to get: a 6.7 GB download that uses 7.5 GB of memory, and writes about 5.9 words a second.

Qwen3 8B 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 Qwen/Qwen3-8B-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

SizeDownloadMemory at 8KGet
Q4_K_M5.0 GB5.8 GBestimatedOpen
Q5_05.7 GB6.5 GBestimatedOpen
Q5_K_M5.9 GB6.7 GBestimatedOpen
Q6_K6.7 GB7.5 GBestimatedOpen
Q8_08.7 GB9.5 GBestimatedOpen

Run It Anywhere

Qwen3 8B is an open model, so any app that runs GGUF files can run it. These take the Q6_K size from Qwen straight from Hugging Face.

  • LM StudioMac, Windows, LinuxOpen
  • JanMac, Windows, LinuxOpen
  • MstyMac, Windows, LinuxOpen
  • OllamaMac, Windows, Linuxollama run hf.co/Qwen/Qwen3-8B-GGUF:Q6_K
  • llama.cppMac, Windows, Linuxllama-server -hf Qwen/Qwen3-8B-GGUF:Q6_K
  • Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/Qwen/Qwen3-8B-GGUF:Q6_K
On a PC Graphics Card
Video memoryLargest size that fitsDownloadMemory at 8K
6 GBQ4_K_M5.0 GB5.8 GB
8 GBQ6_K6.7 GB7.5 GB
10 GBQ8_08.7 GB9.5 GB
11 GBQ8_08.7 GB9.5 GB
12 GBQ8_08.7 GB9.5 GB
16 GBQ8_08.7 GB9.5 GB
20 GBQ8_08.7 GB9.5 GB
24 GBQ8_08.7 GB9.5 GB
32 GBQ8_08.7 GB9.5 GB
48 GBQ8_08.7 GB9.5 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 GBQ6_KRuns great7.5 GB8.6
M6 · 24 GBQ8_0Runs great10 GB8.4
M6 · 32 GBQ8_0Runs great12 GB8.4
M5 · 16 GBQ6_KRuns great7.5 GB7.9
M5 · 24 GBQ8_0Runs great10 GB7.6
M5 · 32 GBQ8_0Runs great12 GB7.6
M5 Pro · 24 GBQ8_0Runs great10 GB15
M5 Pro · 48 GBQ8_0Runs great12 GB15
M5 Pro · 64 GBQ8_0Runs great12 GB15
M5 Max · 36 GBQ8_0Runs great12 GB22
M5 Max · 48 GBQ8_0Runs great12 GB29
M5 Max · 64 GBQ8_0Runs great12 GB29
M5 Max · 128 GBQ8_0Runs great12 GB29
M5 Ultra · 96 GBQ8_0Runs great12 GB56
M5 Ultra · 256 GBQ8_0Runs great12 GB56
M5 Ultra · 512 GBQ8_0Runs great12 GB56
M4 · 16 GBQ6_KRuns great7.5 GB5.9
M4 · 24 GBQ8_0Runs great10 GB5.8
M4 · 32 GBQ8_0Runs great12 GB5.8
M4 Pro · 24 GBQ8_0Runs great10 GB12
M4 Pro · 48 GBQ8_0Runs great12 GB12
M4 Pro · 64 GBQ8_0Runs great12 GB12
M4 Max · 36 GBQ8_0Runs great12 GB18
M4 Max · 48 GBQ8_0Runs great12 GB23
M4 Max · 64 GBQ8_0Runs great12 GB23
M4 Max · 96 GBQ8_0Runs great12 GB18
M4 Max · 128 GBQ8_0Runs great12 GB23
M3 · 8 GBUD-Q3_K_XLRuns well5.1 GB6.3
M3 · 16 GBQ6_KRuns great7.5 GB5
M3 · 24 GBQ8_0Runs great10 GB4.9
M3 Pro · 18 GBUD-Q6_K_XLRuns great8.3 GB6.9
M3 Pro · 36 GBQ8_0Runs great12 GB7.2
M3 Max · 36 GBQ8_0Runs great12 GB14
M3 Max · 48 GBQ8_0Runs great12 GB18
M3 Max · 64 GBQ8_0Runs great12 GB18
M3 Max · 96 GBQ8_0Runs great12 GB14
M3 Max · 128 GBQ8_0Runs great12 GB18
M3 Ultra · 96 GBQ8_0Runs great12 GB32
M3 Ultra · 256 GBQ8_0Runs great12 GB32
M3 Ultra · 512 GBQ8_0Runs great12 GB32
M2 · 8 GBUD-Q3_K_XLRuns well5.1 GB6.3
M2 · 16 GBQ6_KRuns great7.5 GB5
M2 · 24 GBQ8_0Runs great10 GB4.9
M2 Pro · 16 GBQ6_KRuns great7.5 GB9.7
M2 Pro · 32 GBQ8_0Runs great12 GB9.5
M2 Max · 32 GBQ8_0Runs great12 GB18
M2 Max · 64 GBQ8_0Runs great12 GB18
M2 Max · 96 GBQ8_0Runs great12 GB18
M2 Ultra · 64 GBQ8_0Runs great12 GB32
M2 Ultra · 128 GBQ8_0Runs great12 GB32
M2 Ultra · 192 GBQ8_0Runs great12 GB32
M1 · 8 GBUD-Q3_K_XLRuns well5.1 GB4.2
M1 · 16 GBQ6_KRuns great7.5 GB3.4
M1 Pro · 16 GBQ6_KRuns great7.5 GB9.2
M1 Pro · 32 GBQ8_0Runs great12 GB9.1
M1 Max · 32 GBQ8_0Runs great12 GB17
M1 Max · 64 GBQ8_0Runs great12 GB17
M1 Ultra · 64 GBQ8_0Runs great12 GB28
M1 Ultra · 128 GBQ8_0Runs great12 GB28

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

Packaged by

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

Information

Maker
Alibaba Qwen
Released
April 2025
Parameters
8.2B
Longest chat
40K tokens
Reads pictures
No
Licence
apache-2.0
Downloads this month
1,072,491
Likes
617
Sizes
5, 5.0 to 8.7 GB
Kind
Maker's release

Badge for the model card

Shows on Qwen3 8B'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