---
title: Nex-N2-Pro-mlx-8bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/Nex-N2-Pro-mlx-8bit"
md_url: "https://www.modelscope.cn/models/mlx-community/Nex-N2-Pro-mlx-8bit.md"
repository: mlx-community/Nex-N2-Pro-mlx-8bit
last_updated: 2026-06-15
license: apache-2.0
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - qwen3_5_moe
architectures:
  - Qwen3_5MoeForConditionalGeneration
base_model:
  - nex-agi/Nex-N2-Pro
base_model_relation: quantized
parameters: 111.5B
tensor_type:
  - BF16
  - U32
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 14
stars: 0
tags:
  - mlx
  - qwen3_5_moe
  - quantized
  - 8-bit
  - moe
  - distributed
---

# Nex-N2-Pro-mlx-8bit

> Nex-N2-Pro-mlx-8bit - mlx-community 在 ModelScope 开源的模型。8-bit (affine, group-size 64) MLX quantization of nex-agi/Nex-N2-Pro (qwen35moe, 397B-A17B). Produced with mlxlm and verified serving distributed across a 4-node Apple-Silicon JACCL/RDMA cluster…

mlx-community/Nex-N2-Pro-mlx-8bit 是 ModelScope 魔搭社区上的 111.5B 参数text-generation模型，采用 apache-2.0 许可，基于 nex-agi/Nex-N2-Pro 构建。

- **Repository**: mlx-community/Nex-N2-Pro-mlx-8bit
- **License**: apache-2.0
- **Tasks**: text-generation
- **Parameters**: 111.5B
- **Base model**: nex-agi/Nex-N2-Pro
- **Tags**: mlx, qwen3_5_moe, quantized, 8-bit, moe, distributed
- **Downloads**: 14
- **Stars**: 0
- **Last updated**: 2026-06-15

Source: https://www.modelscope.cn/models/mlx-community/Nex-N2-Pro-mlx-8bit

---

# Nex-N2-Pro-mlx-8bit

8-bit (affine, group-size 64) MLX quantization of [nex-agi/Nex-N2-Pro](https://hf.co/nex-agi/Nex-N2-Pro) (qwen3_5_moe, 397B-A17B). Produced with `mlx_lm` and **verified serving distributed across a 4-node Apple-Silicon JACCL/RDMA cluster** (coherent output, ~29 tok/s decode).

## Highlights

- **Router precision preserved.** Per the qwen3_5 quantization predicate, every layer's `mlp.gate` and `mlp.shared_expert_gate` are kept at **8-bit** (the `quantization` map in `config.json` lists all 60 layers) — MoE routing stability is not degraded. `A_log` (GatedDeltaNet decay) stays fp32.
- **Tensor-parallel / pipeline ready.** Runs under `mlx_lm` distributed serving (`mlx.launch --backend jaccl`); validated 4-node and 2-node.
- **Text serving.** `mlx_lm` strips the vision tower at load and serves the text `language_model` (responses carry a separate `reasoning` field — it's a thinking model).

## Use with MLX

```bash
pip install mlx-lm
mlx_lm.generate --model mlx-community/Nex-N2-Pro-mlx-8bit \
  --prompt "Write a Python function to merge two sorted lists." --max-tokens 512
```

Distributed (hostfile with the JACCL RDMA device matrix, per the MLX docs):

```bash
mlx.launch --backend jaccl --hostfile hostfile.json -- \
  python -m mlx_lm server --model mlx-community/Nex-N2-Pro-mlx-8bit --port 8080
```

## Quantization details

| | |
|---|---|
| Method | MLX affine (`mlx_lm`) |
| Bits | 8 |
| Group size | 64 |
| Router / shared-expert gates | kept 8-bit (predicate) |
| Size on disk | ~392 GB (91 shards) |
| Architecture | qwen3_5_moe, 60 layers (45 GatedDeltaNet linear + 15 full-attention), 512 experts, 262K ctx |

Quantized from the full-precision `nex-agi/Nex-N2-Pro` weights. A distillation-aware 4-bit (DWQ) variant is in progress. Apache-2.0, inherited from the base model.