What are the hardware requirements for MiniMax M2?

MiniMax M2 requires approximately 77.7 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 MiniMax M2 and how large is it?

MiniMax M2 is developed by MiniMaxAI, a model with 228.7b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does MiniMax M2 perform on standard benchmarks?

MiniMax M2 has the following published scores: LMSYS ELO 1346. Compare against reference models on this page or on the LLM Explorer leaderboards.

MiniMax M2 by MiniMaxAI — VRAM 77.7GB, 192K context

Name: MiniMax M2
Author: MiniMaxAI

MiniMax M2 is an open-source language model by MiniMaxAI. Features: 228.7b LLM, VRAM: 77.7GB, Context: 192K, License: other, LLM Explorer Score: 0.49, ELO: 1346.

Arxiv:2504.07164 Arxiv:2509.06501 Arxiv:2509.13160 Conversational Custom code Deploy:azure Endpoints compatible Eval-results Fp8 Minimax m2 Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/MiniMaxAI/MiniMax-M2

MiniMax M2 Benchmarks

LMSys ELO 1346 vs 1272 (so35) 5.8%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

MiniMax M2 Parameters and Internals

LLM Name	MiniMax M2
Repository 🤗	https://huggingface.co/MiniMaxAI/MiniMax-M2
Model Size	228.7b
Required VRAM	77.7 GB
Updated	2026-05-17
Maintainer	MiniMaxAI
Model Type	minimax_m2
Model Files	3.7 GB: 0-of-130 1.2 GB: 1-of-130 2.5 GB: 2-of-130 1.2 GB: 3-of-130 2.5 GB: 4-of-130 1.2 GB: 5-of-130 2.5 GB: 6-of-130 1.2 GB: 7-of-130 2.5 GB: 8-of-130 1.2 GB: 9-of-130 2.5 GB: 10-of-130 1.2 GB: 11-of-130 2.5 GB: 12-of-130 1.2 GB: 13-of-130 2.5 GB: 14-of-130 1.2 GB: 15-of-130 2.5 GB: 16-of-130 1.2 GB: 17-of-130 2.5 GB: 18-of-130 1.2 GB: 19-of-130 2.5 GB: 20-of-130 1.2 GB: 21-of-130 2.5 GB: 22-of-130 1.2 GB: 23-of-130 2.5 GB: 24-of-130 1.2 GB: 25-of-130 2.5 GB: 26-of-130 1.2 GB: 27-of-130 2.5 GB: 28-of-130 1.2 GB: 29-of-130 2.5 GB: 30-of-130 1.2 GB: 31-of-130 2.5 GB: 32-of-130 1.2 GB: 33-of-130 2.5 GB: 34-of-130 1.2 GB: 35-of-130 2.5 GB: 36-of-130 1.2 GB: 37-of-130 2.5 GB: 38-of-130 1.2 GB: 39-of-130 2.5 GB: 40-of-130
Model Architecture	MiniMaxM2ForCausalLM
License	other
Context Length	196608
Model Max Length	196608
Transformers Version	4.57.1
Vocabulary Size	200064

Rank the MiniMax M2 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

Quantized Models of the MiniMax M2

Model	Likes	Downloads	VRAM
MiniMax M2 Mlx 8bit Gs32	2	14	206 GB
MiniMax M2 4bit	10	339	128 GB
MiniMax M2 8bit	8	70	223 GB
MiniMax M2 3bit	3	43	99 GB
MiniMax M2 5bit	2	27	155 GB
MiniMax M2 6bit	2	14	185 GB

Best Alternatives to MiniMax M2

Best Alternatives	Context / RAM	Downloads	Likes
MiniMax M2.7	200K / 75.2 GB	923544	1235
List 3.0 Ultra Coder Brain	200K / 75.2 GB	22	10
MiniMax M2.7 BF16	200K / 147.3 GB	18	0
MiniMax M2.5	192K / 74 GB	772099	1502
MiniMax M2.7	192K / 213 GB	247	2
MiniMax M2.7	192K / 75.2 GB	58	18
MiniMax M2.5	192K / 75.2 GB	19	5
MiniMax M2.1	192K / 75.2 GB	54	13
MiniMax M2.1	192K / 77.7 GB	12798	1358
MiniMax M2	192K / 77.7 GB	24	17

Note: green Score (e.g. "73.2") means that the model is better than MiniMaxAI/MiniMax-M2.