---
name: MiniMax-M2.7-REAP-139B-A10B-GGUF
base_model: MiniMaxAI/MiniMax-M2.7
license: other
license_name: modified-mit
license_link: https://hf.co/MiniMaxAI/MiniMax-M2.7/blob/main/LICENSE
pipeline_tag: text-generation
tasks: text-generation
language: en
library_name: llama.cpp
papers: https://arxiv.org/abs/2510.13999
tags:
- Cerebras
- MiniMaxAI
- M2.7
- REAP
- GGUF
- static quantization
---
> [!IMPORTANT]
> Original Model Link : [dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B](https://hf.co/dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B)
>

```
name: MiniMax-M2.7-REAP-139B-A10B-GGUF
base_model: MiniMaxAI/MiniMax-M2.7
license: other
license_name: modified-mit
license_link: https://hf.co/MiniMaxAI/MiniMax-M2.7/blob/main/LICENSE
pipeline_tag: text-generation
tasks: text-generation
language: en
library_name: llama.cpp
papers: https://arxiv.org/abs/2510.13999
tags:
- Cerebras
- MiniMaxAI
- M2.7
- REAP
- GGUF
- static quantization
```

# Description

This is a 230 billion parameter MiniMax M2.7 model with 40% of its experts pruned with REAP (Router-weighted Expert Activation Pruning), then converted to GGUF with llama.cpp and static quantized.

Command sequence using source version of llama.cpp from source and `/opt/homebrew/Cellar/llama.cpp/8940 (78433f606)` llama-quantize:

```sh
hf download dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B
python -m convert_hf_to_gguf ~/.cache/huggingface/...
llama-quantize MiniMax-M2.5-REAP-139B-A10B-BF16.gguf MXFP4_MOE
```