---
title: MiniMax-M3
canonical_url: "https://www.modelscope.cn/models/unsloth/MiniMax-M3"
md_url: "https://www.modelscope.cn/models/unsloth/MiniMax-M3.md"
repository: unsloth/MiniMax-M3
last_updated: 2026-06-12
license: other
pipeline_tag: image-text-to-text
tasks:
- image-text-to-text
model_type:
- minimax_m3_vl
architectures:
- MiniMaxM3SparseForConditionalGeneration
base_model:
- MiniMaxAI/MiniMax-M3
base_model_relation: finetune
parameters: 427.0B
tensor_type:
- BF16
- F32
library_name:
- safetensors
- pytorch
frameworks:
- pytorch
downloads: 64
stars: 0
tags:
- multimodal
- moe
- agent
- coding
- video
---
# MiniMax-M3
> MiniMax-M3 - unsloth 在 ModelScope 开源的模型。Join Our 💬 WeChat | 🧩 Discord community. MiniMax Agent | ⚡️ API | CLI | MiniMax Website 🤗 Hugging Face | 🐙 GitHub | 🤖️ ModelScope | 📄 LICENSE
unsloth/MiniMax-M3 是 ModelScope 魔搭社区上的 427.0B 参数image-text-to-text模型，采用 other 许可，基于 MiniMaxAI/MiniMax-M3 构建。
- **Repository**: unsloth/MiniMax-M3
- **License**: other
- **Tasks**: image-text-to-text
- **Parameters**: 427.0B
- **Base model**: MiniMaxAI/MiniMax-M3
- **Tags**: multimodal, moe, agent, coding, video
- **Downloads**: 64
- **Stars**: 0
- **Last updated**: 2026-06-12
Source: https://www.modelscope.cn/models/unsloth/MiniMax-M3
---
MiniMax-M3 is a native multimodal model with 1M context. It has ~428B parameters and ~23B activated parameters.
**Highlights:**
- **Native Multimodality:** M3 undergoes mixed-modality training from the very first step, enabling deeper semantic fusion across text, image, and video.
- **Context Scaling via Sparse Attention:** M3 introduces MiniMax Sparse Attention (MSA) to improve long context efficiency. M3 delivers 9× prefill and 15× decode speedups compared to M2 at 1M context, reducing per-token compute to 1/20.
- **Coding & Cowork Capability:** M3 achieves frontier-level performance across long-horizon agentic benchmarks, excelling in both coding and cowork.
## Model Details
| | |
| --- | --- |
| Architecture | MoE + MSA (MiniMax Sparse Attention) |
| Total Parameters | ~428B |
| Activated Parameters | ~23B |
| Experts | 128 (4 active per token) |
| Layers | 60 |
| Context Length | 1M tokens |
| Modalities | Text, Image, Video |
| Precision | bfloat16 |
| Transformers | ≥ 4.52.4 (`trust_remote_code=True`) |
| License | [MiniMax Community License](LICENSE) |
## How to Use
- [MiniMax Agent](https://agent.minimax.io/)
- [MiniMax API](https://platform.minimax.io/)
M3 supports two reasoning modes:
- **thinking** — for complex reasoning, agentic tasks, and long-horizon collaboration.
- **non-thinking** — for latency-sensitive scenarios such as chat and code completion.
## Local Deployment
Download the model:
```bash
hf download MiniMaxAI/MiniMax-M3 --local-dir MiniMax-M3
```
We recommend the following inference frameworks (listed alphabetically) to serve the model:
### SGLang
We recommend using [SGLang](https://docs.sglang.io/) to serve MiniMax-M3. Please refer to our [SGLang Deployment Guide](./docs/sglang_deploy_guide.md).
### vLLM
We recommend using [vLLM](https://github.com/vllm-project/vllm) to serve MiniMax-M3. Please refer to our [vLLM Deployment Guide](./docs/vllm_deploy_guide.md).
### Transformers
We recommend using [Transformers](https://github.com/huggingface/transformers) to serve MiniMax-M3. Please refer to our [Transformers Deployment Guide](./docs/transformers_deploy_guide.md).
### ModelScope
You can also get model weights from [ModelScope](https://modelscope.cn/models/MiniMax/MiniMax-M3).
### Inference Parameters
We recommend the following parameters for best performance: `temperature=1.0`, `top_p=0.95`, `top_k=40`. Default system prompt:
```
You are a helpful assistant. Your name is MiniMax-M3 and was built by MiniMax.
```
## Tool Calling Guide
Please refer to our [Tool Calling Guide](./docs/tool_calling_guide.md).
## Contact Us
Contact us at [model@minimax.io](mailto:model@minimax.io).
