]~!b[]~b]system
Your model version is MiniMax-M3, developed by MiniMax. Knowledge cutoff: January 2026. Founded in early 2022, MiniMax is a global AI foundation model company committed to advancing the frontiers of AI towards AGI.

<thinking_instructions>
You have a thinking capability that allows you to reason step by step before responding. When thinking is enabled, wrap your reasoning in <mm:think></mm:think> tags before your response. When thinking is disabled, begin your response directly after the </mm:think> prefix. When thinking is adaptive, decide on your own whether to think for the current turn.
Current thinking mode: adaptive. You are encouraged to think for complex decision-making, multi-step reasoning, or when analyzing function/tool results.
</thinking_instructions>[e~[
]~b]developer
{system_prompt}[e~[
]~b]user
{prompt}[e~[
]~b]ai

Don't know which to choose? Grab Q4_K_M (261.28GB) - usually a good mix of size and performance. Download instructions available here

Available files:

Filename	Quant type	File Size	Split	Description
MiniMax-M3-Q8_0.gguf	Q8_0	453.61GB	true	Extremely high quality, generally unneeded but max available quant.
MiniMax-M3-Q6_K.gguf	Q6_K	369.40GB	true	Very high quality, near perfect, recommended.
MiniMax-M3-Q5_K_M.gguf	Q5_K_M	305.35GB	true	High quality, recommended.
MiniMax-M3-Q5_K_S.gguf	Q5_K_S	295.23GB	true	High quality, recommended.
MiniMax-M3-Q4_1.gguf	Q4_1	268.89GB	true	Legacy format, similar performance to Q4_K_S but with improved tokens/watt on Apple silicon.
MiniMax-M3-Q4_K_M.gguf	Q4_K_M	261.28GB	true	Good quality, default size for most use cases, recommended.
MiniMax-M3-Q4_K_S.gguf	Q4_K_S	251.36GB	true	Slightly lower quality with more space savings, recommended.
MiniMax-M3-Q4_0.gguf	Q4_0	243.64GB	true	Legacy format, kept for compatibility with older tools.
MiniMax-M3-IQ4_NL.gguf	IQ4_NL	242.75GB	true	Similar to IQ4_XS, but slightly larger.
MiniMax-M3-IQ4_XS.gguf	IQ4_XS	229.69GB	true	Decent quality, smaller than Q4_K_S with similar performance, recommended.
MiniMax-M3-Q3_K_XL.gguf	Q3_K_XL	206.05GB	true	Uses Q8_0 for embed and output weights. Lower quality but usable, good for low RAM availability.
MiniMax-M3-IQ3_M.gguf	IQ3_M	205.52GB	true	Medium-low quality, new method with decent performance comparable to Q3_K_M.
MiniMax-M3-Q3_K_L.gguf	Q3_K_L	204.97GB	true	Lower quality but usable, good for low RAM availability.
MiniMax-M3-Q3_K_M.gguf	Q3_K_M	196.61GB	true	Low quality.
MiniMax-M3-IQ3_XS.gguf	IQ3_XS	196.36GB	true	Lower quality, new method with decent performance, slightly better than Q3_K_S.
MiniMax-M3-Q3_K_S.gguf	Q3_K_S	187.25GB	true	Low quality, not recommended.
MiniMax-M3-IQ3_XXS.gguf	IQ3_XXS	180.01GB	true	Lower quality, new method with decent performance, comparable to Q3 quants.
MiniMax-M3-Q2_K_L.gguf	Q2_K_L	153.09GB	true	Uses Q8_0 for embed and output weights. Very low quality but surprisingly usable.
MiniMax-M3-Q2_K.gguf	Q2_K	151.89GB	true	Very low quality but surprisingly usable.
MiniMax-M3-IQ2_M.gguf	IQ2_M	145.58GB	true	Relatively low quality, uses SOTA techniques to be surprisingly usable.
MiniMax-M3-IQ2_S.gguf	IQ2_S	132.14GB	true	Low quality, uses SOTA techniques to be usable.
MiniMax-M3-IQ2_XS.gguf	IQ2_XS	129.52GB	true	Low quality, uses SOTA techniques to be usable.
MiniMax-M3-IQ2_XXS.gguf	IQ2_XXS	116.61GB	true	Very low quality, uses SOTA techniques to be usable.
MiniMax-M3-IQ1_M.gguf	IQ1_M	100.74GB	true	Extremely low quality, not recommended.
MiniMax-M3-IQ1_S.gguf	IQ1_S	90.53GB	true	Extremely low quality, not recommended.

Downloading using the Hugging Face CLI

Click to view download instructions

First, make sure you have the Hugging Face CLI installed:

pip install -U "huggingface_hub[cli]"

The files marked true in the Split column above are stored as multiple parts in a folder. To download all the parts to a local folder, run:

hf download bartowski/MiniMax-M3-GGUF --include "MiniMax-M3-Q8_0/*" --local-dir ./

You can either specify a new local-dir (MiniMax-M3-Q8_0) or download them all in place (./)

How to run

These quants run with llama.cpp - installable in one line via llama.app:

curl -LsSf https://llama.app/install.sh | sh
llama-server -hf bartowski/MiniMax-M3-GGUF:Q4_K_M

llama-server includes a built-in chat web UI, served at http://localhost:8080 by default.

These quants were made with llama.cpp release b10141 - if this model's architecture is newly supported, you'll need that release or newer to run them.

They also work in: LM Studio · koboldcpp · ramalama · Jan AI · Text Generation Web UI · LoLLMs · Atomic Chat

Multimodal

This model supports multimodal input. Alongside the quants, this repo includes the multimodal projector files mmproj-MiniMax-M3-f16.gguf and mmproj-MiniMax-M3-bf16.gguf, which pair with any quant above.

llama.cpp downloads the mmproj automatically when using -hf as shown above; if you're loading files manually, pass it with --mmproj.

imatrix

All quants made using imatrix option with dataset from here. The imatrix is available here: MiniMax-M3-imatrix.gguf.

Embed/output weights

Some of these quants (Q3_K_XL, Q4_K_L etc) are the standard quantization method with the embeddings and output weights quantized to Q8_0 instead of what they would normally default to.

ARM/AVX information

llama.cpp automatically "repacks" weights into an interleaved layout at load time for faster inference on ARM and AVX machines - details in this PR. This once required downloading special Q4_0_4_4/4_8/8_8 files; those are long gone. Online repacking now covers Q4_0, IQ4_NL, and most K-quants, so no special quant choice is needed for CPU inference.

Which file should I choose?

Click here for details

An older (early 2024) but still useful write-up with charts comparing quant performances is provided by Artefact2 here

The first thing to figure out is how big a model you can run. To do this, you'll need to figure out how much RAM and/or VRAM you have.

If you want your model running as FAST as possible, you'll want to fit the whole thing on your GPU's VRAM. Aim for a quant with a file size 1-2GB smaller than your GPU's total VRAM.

If you want the absolute maximum quality, add both your system RAM and your GPU's VRAM together, then similarly grab a quant with a file size 1-2GB Smaller than that total.

Hugging Face can also do this math for you: add your hardware in your Local Apps settings and the model page will show which files fit.

Next, you'll need to decide if you want to use an 'I-quant' or a 'K-quant'.

If you don't want to think too much, grab one of the K-quants. These are in format 'QX_K_X', like Q5_K_M.

If you want to get more into the weeds, you can check out this extremely useful feature chart:

llama.cpp feature matrix

But basically, if you're aiming for below Q4, and you're running cuBLAS (Nvidia) or rocBLAS (AMD), you should look towards the I-quants. These are in format IQX_X, like IQ3_M. These are newer and offer better performance for their size.

These I-quants can also be used on CPU, but will be slower than their K-quant equivalent, so speed vs performance is a tradeoff you'll have to decide.

Credits

Thank you kalomaze and Dampf for assistance in creating the imatrix calibration dataset.

Thank you ZeroWw for the inspiration to experiment with embed/output.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski