Models

Qwen3 235B A22B 2507

Alibaba Qwen · July 2025

Get

Hugging Face

A 235B, 22B active chat model from Alibaba Qwen.

Downloads: 24K
Likes: 141
Parameters: 235B, 22B active
Context: 256K
Licence: apache-2.0

Too big for this Mac

Its smallest size needs 83 GB of memory, more than M4 · 16 GB can give it.

Ratings

KnowledgeMMLU-Pro: 83%

MMLU-Pro: graduate-level questions in 14 subjects. 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.

Qwen3 235B A22B 2507 runs well from a Mac with an M2 Max and 96 GB of memory. 15 of 59 Mac configurations run it well.

Compare with

The nearest models in size.

Sizes

From unsloth/Qwen3-235B-A22B-Instruct-2507-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

Size	Download	Memory at 8K	Get
Q2_K	86 GB	87 GBestimated	Open
Q2_K_L	86 GB	87 GBestimated	Open
UD-Q2_K_XL	89 GB	90 GBestimated	Open
Q3_K_S	101 GB	102 GBestimated	Open
UD-Q3_K_XL	104 GB	105 GBestimated	Open
Q3_K_M	112 GB	113 GBestimated	Open
IQ4_XS	125 GB	126 GBestimated	Open
Q4_0	133 GB	134 GBestimated	Open
Q4_K_S	134 GB	135 GBestimated	Open
UD-Q4_K_XL	134 GB	135 GBestimated	Open
Q4_K_M	142 GB	143 GBestimated	Open
Q4_1	147 GB	148 GBestimated	Open
Q5_K_S	162 GB	163 GBestimated	Open
Q5_K_M	167 GB	168 GBestimated	Open
UD-Q5_K_XL	169 GB	170 GBestimated	Open
Q6_K	193 GB	194 GBestimated	Open
UD-Q6_K_XL	202 GB	203 GBestimated	Open
Q8_0	250 GB	251 GBestimated	Open
UD-Q8_K_XL	274 GB	275 GBestimated	Open
BF16	470 GB	471 GBestimated	Open

Run It Anywhere

Qwen3 235B A22B 2507 is an open model, so any app that runs GGUF files can run it. These take the Q2_K size from unsloth straight from Hugging Face.

LM StudioMac, Windows, LinuxOpen
JanMac, Windows, LinuxOpen
MstyMac, Windows, LinuxOpen
OllamaMac, Windows, Linuxollama run hf.co/unsloth/Qwen3-235B-A22B-Instruct-2507-GGUF:Q2_K
llama.cppMac, Windows, Linuxllama-server -hf unsloth/Qwen3-235B-A22B-Instruct-2507-GGUF:Q2_K
Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/unsloth/Qwen3-235B-A22B-Instruct-2507-GGUF:Q2_K

On a PC Graphics Card

Video memory	Largest size that fits	Memory at 8K
6 GB	None worth running	needs 90 GB
8 GB	None worth running	needs 90 GB
10 GB	None worth running	needs 90 GB
11 GB	None worth running	needs 90 GB
12 GB	None worth running	needs 90 GB
16 GB	None worth running	needs 90 GB
20 GB	None worth running	needs 90 GB
24 GB	None worth running	needs 90 GB
32 GB	None worth running	needs 90 GB
48 GB	None worth running	needs 90 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

Mac	Size	Fit	Memory	Writes
M6 · 16 GB	IQ1_M	Too big	83 GB
M6 · 24 GB	IQ1_M	Too big	83 GB
M6 · 32 GB	IQ1_M	Too big	83 GB
M5 · 16 GB	IQ1_M	Too big	83 GB
M5 · 24 GB	IQ1_M	Too big	83 GB
M5 · 32 GB	IQ1_M	Too big	83 GB
M5 Pro · 24 GB	IQ1_M	Too big	83 GB
M5 Pro · 48 GB	IQ1_M	Too big	83 GB
M5 Pro · 64 GB	IQ1_M	Too big	83 GB
M5 Max · 36 GB	IQ1_M	Too big	83 GB
M5 Max · 48 GB	IQ1_M	Too big	83 GB
M5 Max · 64 GB	IQ1_M	Too big	83 GB
M5 Max · 128 GB	IQ3_XS	Runs well	101 GB	14
M5 Ultra · 96 GB	IQ2_S	Runs well	68 GB	30
M5 Ultra · 256 GB	UD-Q5_K_XL	Runs great	173 GB	19
M5 Ultra · 512 GB	Q8_0	Runs great	279 GB	17
M4 · 16 GB	IQ1_M	Too big	83 GB
M4 · 24 GB	IQ1_M	Too big	83 GB
M4 · 32 GB	IQ1_M	Too big	83 GB
M4 Pro · 24 GB	IQ1_M	Too big	83 GB
M4 Pro · 48 GB	IQ1_M	Too big	83 GB
M4 Pro · 64 GB	IQ1_M	Too big	83 GB
M4 Max · 36 GB	IQ1_M	Too big	83 GB
M4 Max · 48 GB	IQ1_M	Too big	83 GB
M4 Max · 64 GB	IQ1_M	Too big	83 GB
M4 Max · 96 GB	IQ2_S	Runs well	68 GB	9.6
M4 Max · 128 GB	IQ3_XS	Runs well	101 GB	10
M3 · 8 GB	IQ1_M	Too big	83 GB
M3 · 16 GB	IQ1_M	Too big	83 GB
M3 · 24 GB	IQ1_M	Too big	83 GB
M3 Pro · 18 GB	IQ1_M	Too big	83 GB
M3 Pro · 36 GB	IQ1_M	Too big	83 GB
M3 Max · 36 GB	IQ1_M	Too big	83 GB
M3 Max · 48 GB	IQ1_M	Too big	83 GB
M3 Max · 64 GB	IQ1_M	Too big	83 GB
M3 Max · 96 GB	IQ2_S	Runs well	68 GB	7.9
M3 Max · 128 GB	IQ3_XS	Runs well	101 GB	8.4
M3 Ultra · 96 GB	IQ2_S	Runs well	68 GB	16
M3 Ultra · 256 GB	UD-Q5_K_XL	Runs great	173 GB	11
M3 Ultra · 512 GB	Q8_0	Runs great	279 GB	9.8
M2 · 8 GB	IQ1_M	Too big	83 GB
M2 · 16 GB	IQ1_M	Too big	83 GB
M2 · 24 GB	IQ1_M	Too big	83 GB
M2 Pro · 16 GB	IQ1_M	Too big	83 GB
M2 Pro · 32 GB	IQ1_M	Too big	83 GB
M2 Max · 32 GB	IQ1_M	Too big	83 GB
M2 Max · 64 GB	IQ1_M	Too big	83 GB
M2 Max · 96 GB	IQ2_S	Runs well	68 GB	9.8
M2 Ultra · 64 GB	IQ1_M	Too big	83 GB
M2 Ultra · 128 GB	IQ3_XS	Runs well	101 GB	14
M2 Ultra · 192 GB	IQ4_XS	Runs great	129 GB	12
M1 · 8 GB	IQ1_M	Too big	83 GB
M1 · 16 GB	IQ1_M	Too big	83 GB
M1 Pro · 16 GB	IQ1_M	Too big	83 GB
M1 Pro · 32 GB	IQ1_M	Too big	83 GB
M1 Max · 32 GB	IQ1_M	Too big	83 GB
M1 Max · 64 GB	IQ1_M	Too big	83 GB
M1 Ultra · 64 GB	IQ1_M	Too big	83 GB
M1 Ultra · 128 GB	IQ3_XS	Runs well	101 GB	11

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: Alibaba Qwen
Released: July 2025
Parameters: 235B, 22B active
Longest chat: 256K tokens
Reads pictures: No
Licence: apache-2.0
Downloads this month: 24,040
Likes: 141
Sizes: 20, 86 to 470 GB
Kind: Maker's release
Source: Qwen/Qwen3-235B-A22B-Instruct-2507

Badge for the model card

Shows on Qwen3 235B A22B 2507's Hugging Face card which Macs run it well, linked to this page. It updates every morning.

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.