What are the hardware requirements for M51Lab MiniMax M2.7 REAP 139B A10B?

M51Lab MiniMax M2.7 REAP 139B A10B requires approximately 205 GB of VRAM and supports a context window of 192K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed M51Lab MiniMax M2.7 REAP 139B A10B and how large is it?

M51Lab MiniMax M2.7 REAP 139B A10B is developed by dervig, a model with 139b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate M51Lab MiniMax M2.7 REAP 139B A10B?

M51Lab MiniMax M2.7 REAP 139B A10B is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

M51Lab MiniMax M2.7 REAP 139B A10B by dervig — VRAM 205GB, 192K context

Name: M51Lab MiniMax M2.7 REAP 139B A10B
Author: dervig

M51Lab MiniMax M2.7 REAP 139B A10B is an open-source language model by dervig. Features: 139b LLM, VRAM: 205GB, Context: 192K, License: other, LLM Explorer Score: 0.24.

Arxiv:2510.13999 Base model:finetune:minimaxai/... Base model:minimaxai/minimax-m... Conversational Custom code Endpoints compatible Minimax Minimax m2 Moe Pruning Reap Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B

M51Lab MiniMax M2.7 REAP 139B A10B Parameters and Internals

LLM Name	M51Lab MiniMax M2.7 REAP 139B A10B
Repository 🤗	https://huggingface.co/dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B
Base Model(s)	MiniMaxAI/MiniMax-M2.7 MiniMaxAI/MiniMax-M2.7
Model Size	139b
Required VRAM	205 GB
Updated	2026-07-26
Maintainer	dervig
Model Type	minimax_m2
Model Files	5.0 GB: 1-of-56 5.0 GB: 2-of-56 5.0 GB: 3-of-56 5.0 GB: 4-of-56 5.0 GB: 5-of-56 5.0 GB: 6-of-56 5.0 GB: 7-of-56 5.0 GB: 8-of-56 5.0 GB: 9-of-56 5.0 GB: 10-of-56 5.0 GB: 11-of-56 5.0 GB: 12-of-56 5.0 GB: 13-of-56 5.0 GB: 14-of-56 5.0 GB: 15-of-56 5.0 GB: 16-of-56 5.0 GB: 17-of-56 5.0 GB: 18-of-56 5.0 GB: 19-of-56 5.0 GB: 20-of-56 5.0 GB: 21-of-56 5.0 GB: 22-of-56 5.0 GB: 23-of-56 5.0 GB: 24-of-56 5.0 GB: 25-of-56 5.0 GB: 26-of-56 5.0 GB: 27-of-56 5.0 GB: 28-of-56 5.0 GB: 29-of-56 5.0 GB: 30-of-56 5.0 GB: 31-of-56 5.0 GB: 32-of-56 5.0 GB: 33-of-56 5.0 GB: 34-of-56 5.0 GB: 35-of-56 5.0 GB: 36-of-56 5.0 GB: 37-of-56 5.0 GB: 38-of-56 5.0 GB: 39-of-56 5.0 GB: 40-of-56 5.0 GB: 41-of-56
Model Architecture	MiniMaxM2ForCausalLM
License	other
Context Length	196608
Model Max Length	196608
Transformers Version	4.55.0
Vocabulary Size	200064
Torch Data Type	bfloat16

Rank the M51Lab MiniMax M2.7 REAP 139B A10B capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to M51Lab MiniMax M2.7 REAP 139B A10B

Best Alternatives	Context / RAM	Downloads	Likes
...iMax M2.7 REAP 139B A10B NVFP4	192K / 80.1 GB	43	5
MiniMax M2.5 REAP 139B A10B	192K / 141.2 GB	68	29
MiniMax M2.1 139B	192K / 140.5 GB	29	10
MiniMax M2 REAP 139B A10B	192K / 140.5 GB	79	18

Note: green Score (e.g. "73.2") means that the model is better than dervig/m51Lab-MiniMax-M2.7-REAP-139B-A10B.