Models

Mistral Medium 3.5 128B

Mistral AI · March 2026

Get

Hugging Face

A 125B chat model from Mistral AI that reads pictures as well as text.

Downloads: 19K
Likes: 101
Parameters: 125B
Context: 256K
Licence: other

Too big for this Mac

Its smallest size needs 88 GB of memory, more than M4 · 16 GB can give it.

Ratings

Software fixesSWE-bench Verified: 77.6%

SWE-bench Verified: real GitHub issues fixed so the project's tests pass. 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.

Mistral Medium 3.5 128B runs well from a Mac with an M1 Max and 64 GB of memory. 24 of 59 Mac configurations run it well.

Compare with

The nearest models in size.

Sizes

From unsloth/Mistral-Medium-3.5-128B-GGUF on Hugging Face. A smaller size is a smaller download and a little less accurate.

Size	Download	Memory at 8K	Get
UD-IQ2_XXS	35 GB	37 GBestimated	Open
UD-IQ2_M	44 GB	46 GBestimated	Open
Q2_K	47 GB	48 GBestimated	Open
Q2_K_L	47 GB	49 GBestimated	Open
UD-Q2_K_XL	48 GB	50 GBestimated	Open
UD-IQ3_XXS	49 GB	51 GBestimated	Open
Q3_K_S	54 GB	56 GBestimated	Open
Q3_K_M	61 GB	62 GBestimated	Open
UD-Q3_K_XL	62 GB	64 GBestimated	Open
IQ4_XS	67 GB	69 GBestimated	Open
IQ4_NL	71 GB	73 GBestimated	Open
Q4_0	71 GB	73 GBestimated	Open
Q4_K_S	71 GB	73 GBestimated	Open
Q4_K_M	75 GB	77 GBestimated	Open
UD-Q4_K_XL	76 GB	77 GBestimated	Open
Q4_1	78 GB	80 GBestimated	Open
Q5_K_S	86 GB	88 GBestimated	Open
Q5_K_M	88 GB	90 GBestimated	Open
UD-Q5_K_XL	88 GB	90 GBestimated	Open
Q6_K	103 GB	104 GBestimated	Open
UD-Q6_K_XL	109 GB	111 GBestimated	Open
Q8_0	133 GB	135 GBestimated	Open
UD-Q8_K_XL	145 GB	147 GBestimated	Open
BF16	250 GB	252 GBestimated	Open

Run It Anywhere

Mistral Medium 3.5 128B is an open model, so any app that runs GGUF files can run it. These take the UD-IQ2_XXS size from unsloth straight from Hugging Face.

LM StudioMac, Windows, LinuxOpen
JanMac, Windows, LinuxOpen
MstyMac, Windows, LinuxOpen
OllamaMac, Windows, Linuxollama run hf.co/unsloth/Mistral-Medium-3.5-128B-GGUF:UD-IQ2_XXS
llama.cppMac, Windows, Linuxllama-server -hf unsloth/Mistral-Medium-3.5-128B-GGUF:UD-IQ2_XXS
Docker Model RunnerMac, Windows, Linuxdocker model run hf.co/unsloth/Mistral-Medium-3.5-128B-GGUF:UD-IQ2_XXS

On a PC Graphics Card

Video memory	Largest size that fits	Download	Memory at 8K
6 GB	None worth running		needs 46 GB
8 GB	None worth running		needs 46 GB
10 GB	None worth running		needs 46 GB
11 GB	None worth running		needs 46 GB
12 GB	None worth running		needs 46 GB
16 GB	None worth running		needs 46 GB
20 GB	None worth running		needs 46 GB
24 GB	None worth running		needs 46 GB
32 GB	None worth running		needs 46 GB
48 GB	UD-IQ2_M	44 GB	46 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	UD-IQ2_XXS	Too big	88 GB
M6 · 24 GB	UD-IQ2_XXS	Too big	88 GB
M6 · 32 GB	UD-IQ2_XXS	Too big	88 GB
M5 · 16 GB	UD-IQ2_XXS	Too big	88 GB
M5 · 24 GB	UD-IQ2_XXS	Too big	88 GB
M5 · 32 GB	UD-IQ2_XXS	Too big	88 GB
M5 Pro · 24 GB	UD-IQ2_XXS	Too big	88 GB
M5 Pro · 48 GB	IQ1_S	Tight fit	37 GB	3.7
M5 Pro · 64 GB	IQ2_XXS	Runs well	42 GB	3.3
M5 Max · 36 GB	UD-IQ2_XXS	Too big	88 GB
M5 Max · 48 GB	IQ1_S	Tight fit	37 GB	7.3
M5 Max · 64 GB	IQ2_XXS	Runs well	42 GB	6.6
M5 Max · 128 GB	UD-Q3_K_XL	Runs great	66 GB	4.6
M5 Ultra · 96 GB	UD-Q3_K_XL	Runs well	66 GB	9.1
M5 Ultra · 256 GB	Q8_0	Runs great	159 GB	5.7
M5 Ultra · 512 GB	Q8_0	Runs great	186 GB	5.7
M4 · 16 GB	UD-IQ2_XXS	Too big	88 GB
M4 · 24 GB	UD-IQ2_XXS	Too big	88 GB
M4 · 32 GB	UD-IQ2_XXS	Too big	88 GB
M4 Pro · 24 GB	UD-IQ2_XXS	Too big	88 GB
M4 Pro · 48 GB	IQ1_S	Tight fit	37 GB	3.2
M4 Pro · 64 GB	IQ2_XXS	Runs well	42 GB	2.9
M4 Max · 36 GB	UD-IQ2_XXS	Too big	88 GB
M4 Max · 48 GB	IQ1_S	Tight fit	37 GB	6.1
M4 Max · 64 GB	IQ2_XXS	Runs well	42 GB	5.6
M4 Max · 96 GB	UD-Q3_K_XL	Runs well	66 GB	3.1
M4 Max · 128 GB	UD-Q3_K_XL	Runs great	66 GB	4
M3 · 8 GB	UD-IQ2_XXS	Too big	88 GB
M3 · 16 GB	UD-IQ2_XXS	Too big	88 GB
M3 · 24 GB	UD-IQ2_XXS	Too big	88 GB
M3 Pro · 18 GB	UD-IQ2_XXS	Too big	88 GB
M3 Pro · 36 GB	UD-IQ2_XXS	Too big	88 GB
M3 Max · 36 GB	UD-IQ2_XXS	Too big	88 GB
M3 Max · 48 GB	IQ1_S	Tight fit	37 GB	4.7
M3 Max · 64 GB	IQ2_XXS	Runs well	42 GB	4.3
M3 Max · 96 GB	UD-Q3_K_XL	Runs well	66 GB	2.3
M3 Max · 128 GB	UD-Q3_K_XL	Runs great	66 GB	3
M3 Ultra · 96 GB	UD-Q3_K_XL	Runs well	66 GB	5.8
M3 Ultra · 256 GB	Q8_0	Runs great	159 GB	3.6
M3 Ultra · 512 GB	Q8_0	Runs great	186 GB	3.6
M2 · 8 GB	UD-IQ2_XXS	Too big	88 GB
M2 · 16 GB	UD-IQ2_XXS	Too big	88 GB
M2 · 24 GB	UD-IQ2_XXS	Too big	88 GB
M2 Pro · 16 GB	UD-IQ2_XXS	Too big	88 GB
M2 Pro · 32 GB	UD-IQ2_XXS	Too big	88 GB
M2 Max · 32 GB	UD-IQ2_XXS	Too big	88 GB
M2 Max · 64 GB	IQ2_XXS	Runs well	42 GB	4.2
M2 Max · 96 GB	UD-Q3_K_XL	Runs well	66 GB	3
M2 Ultra · 64 GB	IQ2_XXS	Runs well	42 GB	7.9
M2 Ultra · 128 GB	UD-Q3_K_XL	Runs great	66 GB	5.8
M2 Ultra · 192 GB	UD-Q6_K_XL	Runs great	122 GB	3.7
M1 · 8 GB	UD-IQ2_XXS	Too big	88 GB
M1 · 16 GB	UD-IQ2_XXS	Too big	88 GB
M1 Pro · 16 GB	UD-IQ2_XXS	Too big	88 GB
M1 Pro · 32 GB	UD-IQ2_XXS	Too big	88 GB
M1 Max · 32 GB	UD-IQ2_XXS	Too big	88 GB
M1 Max · 64 GB	IQ2_XXS	Runs well	42 GB	4
M1 Ultra · 64 GB	IQ2_XXS	Runs well	42 GB	7.2
M1 Ultra · 128 GB	UD-Q3_K_XL	Runs great	66 GB	5.4

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: Mistral AI
Released: March 2026
Parameters: 125B
Longest chat: 256K tokens
Reads pictures: Yes
Licence: other
Downloads this month: 18,507
Likes: 101
Sizes: 24, 35 to 250 GB
Kind: Maker's release
Source: mistralai/Mistral-Medium-3.5-128B

Badge for the model card

Shows on Mistral Medium 3.5 128B'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.