vultr/MiniMax-M2.7-GGUF

MiniMax M2.7, GGUF

MiniMax M2.7 is a large-scale Mixture-of-Experts (MoE) language model developed by MiniMax for coding, agentic workflows, and enterprise automation tasks. It is optimized for autonomous agent teams, iterative self-improvement, and complex real-world task execution.

This repository holds Vultr's GGUF quantizations of it. They are hybrid quantizations: rather than applying one precision to the whole network, each layer's precision is chosen from a measured importance score, so the bits go where they change the output most.

Model Overview

Attribute	Value
Base model	MiniMaxAI/MiniMax-M2.7
Architecture	minimax-m2, Mixture-of-Experts
Parameters	229B
Active parameters	approximately 10B
Layers	62
Hidden size	3,072
Attention heads	48
Experts	256, 8 selected per token
Context length	up to 200K tokens
Vocabulary	200,064
License	Modified MIT

Attribute	Value
Base model	MiniMaxAI/MiniMax-M2.7
Architecture	minimax-m2, Mixture-of-Experts
Parameters	229B
Active parameters	approximately 10B
Layers	62
Hidden size	3,072
Attention heads	48
Experts	256, 8 selected per token
Context length	up to 200K tokens
Vocabulary	200,064
License	Modified MIT

Quantizations

File	Type	Size	KLD vs bf16
MiniMax-M2.7-Q5_K_M-*.gguf (2 parts)	Q5_K_M	178.51 GB	0.0386
MiniMax-M2.7-Q4_K_M-*.gguf (2 parts)	Q4_K_M	152.07 GB	0.0556
MiniMax-M2.7-IQ3_M.gguf	IQ3_M	110.62 GB	0.1367
MiniMax-M2.7-IQ3_S.gguf	IQ3_S	109.34 GB	0.1379

File	Type	Size	KLD vs bf16
MiniMax-M2.7-Q5_K_M-*.gguf (2 parts)	Q5_K_M	178.51 GB	0.0386
MiniMax-M2.7-Q4_K_M-*.gguf (2 parts)	Q4_K_M	152.07 GB	0.0556
MiniMax-M2.7-IQ3_M.gguf	IQ3_M	110.62 GB	0.1367
MiniMax-M2.7-IQ3_S.gguf	IQ3_S	109.34 GB	0.1379

KL divergence measures how far a quantized model's next-token probabilities drift from those of the original bf16 model, averaged over every token of the evaluation text. The lower the value, the closer the quantized model is to the original; zero would mean identical output.

Evaluation

Every variant in this repository was evaluated against the original bf16 model for perplexity, KL divergence, top-token agreement, HellaSwag and WinoGrande, under identical conditions. The full results are maintained on the model's page in the Vultr Model Library:

https://docs.vultr.com/models/minimax-m2/minimax-m2.7?tab=quantized

That page is the reference for these files. The summary above is an excerpt from it.

Run

llama-server -hf vultr/MiniMax-M2.7-GGUF:Q4_K_M

The model's chat template is embedded in every file, so no extra prompt formatting is needed.

License

These files are derived from MiniMaxAI/MiniMax-M2.7 and are distributed under the same license, Modified MIT. The model is the work of MiniMax; Vultr produced the quantizations.