---
title: MiniMax-M3.1-Flash-oQ3.5e-mtp
canonical_url: "https://www.modelscope.cn/models/mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp"
md_url: "https://www.modelscope.cn/models/mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp.md"
repository: mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp
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:
  - MiniMaxAI/MiniMax-M3.1-Flash
base_model_relation: quantized
parameters: 64.1B
tensor_type:
  - BF16
  - U8
  - U32
  - F32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 15
stars: 0
tags:
  - mlx
  - oq
  - quantized
  - minimax_m3_vl
  - mtp
---

# MiniMax-M3.1-Flash-oQ3.5e-mtp

> MiniMax-M3.1-Flash-oQ3.5e-mtp - mlx-community 在 ModelScope 开源的模型。MiniMax-M3.1-Flash-oQ3.5e-mtp

mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp 是 ModelScope 魔搭社区上的 64.1B 参数image-text-to-text模型，采用 other 许可，基于 MiniMaxAI/MiniMax-M3.1-Flash 构建。

- **Repository**: mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp
- **License**: other
- **Tasks**: image-text-to-text
- **Parameters**: 64.1B
- **Base model**: MiniMaxAI/MiniMax-M3.1-Flash
- **Tags**: mlx, oq, quantized, minimax_m3_vl, mtp
- **Downloads**: 15
- **Stars**: 0
- **Last updated**: 2026-10-23

Source: https://www.modelscope.cn/models/mlx-community/MiniMax-M3.1-Flash-oQ3.5e-mtp

---

# MiniMax-M3.1-Flash-oQ3.5e-mtp

This model was quantized using [oQ](https://github.com/jundot/omlx) (oMLX v0.7.3.dev1) mixed-precision quantization.

A mixed-precision MLX checkpoint of [MiniMaxAI/MiniMax-M3.1-Flash](https://huggingface.co/MiniMaxAI/MiniMax-M3.1-Flash), a native multimodal mixture-of-experts model for coding, agentic workflows, computer use and cybersecurity with about 428B total and 23B activated parameters. It takes text, image and video inputs with a context window of up to 1M tokens, and uses MiniMax Sparse Attention throughout the model. It includes the language model, the vision encoder and the bundled DSpark draft in `dspark/`, which oMLX Lightning MTP uses for speculative decoding.

The effective average is 3.954 bits/weight for the language model and vision encoder, or 3.978 bits/weight including the DSpark draft.

## Quantization and Bit Distribution

oQ3.5e keeps the routed expert down projections in their source NVFP4 format and requantizes the expert gate and up projections to affine 3-bit. It keeps MXFP8 for the shared experts, the sparse attention indexer and about a third of the attention projections, and stores the remaining projections, the embeddings and the vocabulary head as affine 3- to 8-bit. The router and the DSpark draft keep their source precision.

| Storage format | Logical weights | Share of total | Tensor storage | Effective bits/weight |
|---|---|---|---|---|
| Affine 3-bit, group 64 | 275.604B | 64.536% | 112.296 GiB | 3.50 |
| Affine 4-bit, group 64 | 2.378B | 0.557% | 1.246 GiB | 4.50 |
| NVFP4, group 16 | 137.707B | 32.246% | 72.141 GiB | 4.50 |
| Affine 5-bit, group 64 | 2.460B | 0.576% | 1.575 GiB | 5.50 |
| Affine 6-bit, group 64 | 0.226B | 0.053% | 0.171 GiB | 6.50 |
| Affine 8-bit, group 64 | 2.458B | 0.576% | 2.433 GiB | 8.50 |
| MXFP8, group 32 | 5.496B | 1.287% | 5.278 GiB | 8.25 |
| BF16 | 0.678B | 0.159% | 1.262 GiB | 16.00 |
| FP32 | 0.045B | 0.011% | 0.167 GiB | 32.00 |
| **Total** | **427.052B** | **100%** | **196.569 GiB** | **3.954** |

## Storage Breakdown

| Component | Logical weights | Storage |
|---|---|---|
| Language model | 426.375B | 209.710 GB / 195.308 GiB |
| Vision | 0.677B | 1.354 GB / 1.261 GiB |
| DSpark draft (MTP) | 2.011B | 2.302 GB / 2.144 GiB |
| **Total** | **429.063B** | **213.366 GB / 198.712 GiB** |

## License

MiniMax Community License, following the source model. See `LICENSE`. Commercial use has additional conditions in the license, including displaying "Built with MiniMax M3.1-Flash" and obtaining prior written authorization from MiniMax.