---
title: MiniMax-M3.1-Flash
canonical_url: "https://www.modelscope.cn/models/MiniMax/MiniMax-M3.1-Flash"
md_url: "https://www.modelscope.cn/models/MiniMax/MiniMax-M3.1-Flash.md"
repository: MiniMax/MiniMax-M3.1-Flash
chinese_name: MiniMax-M3.1-Flash
last_updated: 2026-10-23
license: other
pipeline_tag: image-text-to-text
tasks:
- image-text-to-text
model_type:
- minimax_m3_vl
architectures:
- MiniMaxM3SparseForConditionalGeneration
base_model_relation: quantized
parameters: 220.5B
tensor_type:
- U8
- F32
- F8_E4M3
- BF16
library_name:
- safetensors
- pytorch
frameworks:
- pytorch
downloads: 91
stars: 7
tags:
- multimodal
- moe
- agent
- coding
- video
- nvfp4
---
# MiniMax-M3.1-Flash
> MiniMax-M3.1-Flash - MiniMax 在 ModelScope 开源的模型。MiniMax-M3.1-Flash is a native multimodal mixture-of-experts model for coding, agentic workflows, computer use, and cybersecurity. It supports text, image, and video inputs with a context window of up to 1M…
MiniMax/MiniMax-M3.1-Flash 是 ModelScope 魔搭社区上的 220.5B 参数image-text-to-text模型，采用 other 许可。
- **Repository**: MiniMax/MiniMax-M3.1-Flash
- **License**: other
- **Tasks**: image-text-to-text
- **Parameters**: 220.5B
- **Tags**: multimodal, moe, agent, coding, video, nvfp4
- **Downloads**: 91
- **Stars**: 7
- **Last updated**: 2026-10-23
Source: https://www.modelscope.cn/models/MiniMax/MiniMax-M3.1-Flash
---
简体中文
# MiniMax-M3.1-Flash
**MiniMax-M3.1-Flash** is a native multimodal mixture-of-experts model for coding, agentic workflows, computer use, and cybersecurity. It supports text, image, and video inputs with a context window of up to **1M tokens**.
Building on MiniMax-M3, M3.1-Flash improves task performance and inference efficiency with sparse attention throughout the model, quantization-aware training, and DSpark speculative decoding.
**Highlights:**
- **Agentic performance:** A generational leap in agentic tasks, coding, and cybersecurity — 76.2 on BankerToolBench, 73.7% on DeepSWE v1.1, 61.7 on OSWorld V2, 29.6 on Agents' Last Exam, and 82.2 on CyberGym.
- **Efficient by design:** Sparse attention throughout, QAT NVFP4, and DSpark speculative decoding improve inference efficiency.
## Model at a Glance
- **Architecture:** Native multimodal mixture-of-experts (MoE)
- **Parameters:** Approximately 428B total; 23B activated per token
- **Inputs / output:** Text, image, and video inputs; text output
- **Context window:** Up to 1,000,000 tokens
- **Attention:** [MiniMax Sparse Attention](https://arxiv.org/abs/2606.13392) (MSA) throughout the model
- **Weight format:** Quantization-aware training (QAT), NVFP4
- **Speculative decoding:** DSpark draft head included
## Benchmarks
| Benchmark |
MiniMax M3.1 Flash |
DeepSeek V4.1 Flash-1017 |
DeepSeek V4.1 Flash |
Kimi K3 |
GLM-5.3 |
GPT-5.6 Sol |
Opus 5 |
Automation-Bench v1.0.6 |
62.8 |
60.5 |
54.8 |
46.7 |
48.8 |
45.8 |
50.3 |
| BankerToolBench |
76.2 |
— |
73.7 |
69.1 |
74.5 |
74.6 |
76.2 |
| GDP.pdf AA |
20.8 |
— |
12.8 |
22.0 |
11.2 |
27.2 |
21.6 |
| Terminal Bench 4.0 |
38.1 |
42.3 |
31.2 |
12.6 |
37.9 |
39.9 |
51.8 |
| LOCA-Bench 256K |
60.3 |
— |
— |
49.7 |
50.3 |
60.0 |
53.7 |
Agents’ Last Exam |
29.6 |
— |
— |
28.3 |
— |
30.6 |
32.2 |
| OSWorld 2.0 |
61.7 |
— |
54.3 |
58.3 |
— |
65.7 |
70.6 |
PostTrain Bench v1.1 |
39.3 |
— |
30.5 |
36.6 |
37.5 |
36.2 |
40.4 |
| DeepSWE v1.1 |
73.7 |
77.0 |
74.2 |
67.5 |
66.9 |
73.0 |
74.0 |
| NL2Repo-Bench |
61.2 |
68.2 |
65.4 |
58.0 |
58.0 |
56.8 |
75.3 |
| SWE Refactor Bench |
23.0 |
— |
11.8 |
19.5 |
10.4 |
28.5 |
31.0 |
CyberGym |
82.2 |
— |
— |
80.0 |
84.5 |
83.6 |
— |
BabyVision (with tools) |
91.0 |
92.5 |
89.6 |
85.7 |
— |
88.9 |
94.1 |
| HLE (without tools) |
43.4 |
42.6 |
39.2 |
43.5 |
42.3 |
49.5 |
55.6 |
| CharXiv RQ |
82.5 |
— |
69.0 |
84.8 |
— |
85.2 |
83.7 |
Evaluation Notes
- **Automation-Bench v1.0.6:** 600 publicly released tasks.
- **BankerToolBench:** OpenCode 1.17.18; MiniMax-M2.7 judge.
- **GDP.pdf AA:** Taken directly from Artificial Analysis.
- **Terminal Bench 4.0:** Codex 0.149.1.
- **LOCA-Bench 256K:** Official implementation.
- **Agents’ Last Exam:** Official Overall setting; full-set results from the official ALE leaderboard only. Blank = no official result (CLI-only excluded), not zero.
- **OSWorld 2.0:** 0808 Full release; highest reported per model at release, except DeepSeek V4.1 (in-house reproduction).
- **PostTrain Bench v1.1:** Claude Code 2.1.251, 512K context; otherwise official setup.
- **DeepSWE v1.1:** mini-SWE-agent 2.4.5, 512K context.
- **NL2Repo-Bench:** Claude Code 2.1.233, 512K context; reward hacking = 0; two inconsistent tasks corrected (scores use the corrected set).
- **SWE Refactor Bench:** Claude Code 2.1.220, 200K context; otherwise official setup.
- **CyberGym:** Claude Code 2.1.263, 1M context.
- **BabyVision (with tools):** Claude Code 2.1.285; min image side 1,500 px.
- **HLE (without tools):** Text-only subset (2,158 questions).
- **CharXiv RQ:** GPT-5.5 judge.
Results are for the official NVFP4 weights; third-party quantizations may differ.
## What’s New in M3.1-Flash
M3.1-Flash retains M3’s core mixture-of-experts architecture, with three main changes:
- **Sparse attention throughout:** the first three full-attention layers are replaced with sparse attention.
- **QAT NVFP4:** routed MoE experts move from MXFP8 to NVFP4 W4A4, excluding the shared expert. The sparse-attention indexer and main branch move from Q8KV8 to Q8KV4.
- **DSpark speculative decoding:** DSpark replaces the earlier EAGLE-like MTP approach. The release includes the draft head for use with compatible serving implementations.
## How to Use
### Local Deployment
Download the model:
```bash
hf download MiniMaxAI/MiniMax-M3.1-Flash --local-dir MiniMax-M3.1-Flash
```
We recommend the following inference frameworks to serve the model:
- [SGLang](https://docs.sglang.io/) — see [SGLang cookbook](https://docs.sglang.io/cookbook/autoregressive/MiniMax/MiniMax-M3.1).
- [vLLM](https://github.com/vllm-project/vllm) — see [vLLM recipe](https://recipes.vllm.ai/MiniMaxAI/MiniMax-M3.1-Flash).
Additional integrations in progress: Transformers, TokenSpeed, KTransformers, unsloth, and ATOM.
## Inference Settings
Recommended sampling settings: `temperature=1.0` and `top_p=0.95`.
For reasoning, set `reasoning_effort` in the chat template: `low`, `medium`, `high`, `xhigh`, or `max`.
The model supports a context window of up to 1M tokens. Usable context length and throughput depend on GPU memory and the serving configuration.
## Contact Us
Contact us at [model@minimax.io](mailto:model@minimax.io).
