---
title: Mistral-Large-4-1T-A52B-Text-4bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/Mistral-Large-4-1T-A52B-Text-4bit"
md_url: "https://www.modelscope.cn/models/mlx-community/Mistral-Large-4-1T-A52B-Text-4bit.md"
repository: mlx-community/Mistral-Large-4-1T-A52B-Text-4bit
last_updated: 2026-10-29
license: other
pipeline_tag: text-generation
tasks:
  - text-generation
model_type:
  - mistral_large4
architectures:
  - MistralLarge4ForConditionalGeneration
base_model:
  - mistralai/Mistral-Large-4-1T-A52B
base_model_relation: quantized
parameters: 164.0B
tensor_type:
  - U32
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 5
stars: 1
tags:
  - mlx
---

# Mistral-Large-4-1T-A52B-Text-4bit

> Mistral-Large-4-1T-A52B-Text-4bit - mlx-community 在 ModelScope 开源的模型。mlx-community/Mistral-Large-4-1T-A52B-Text-4bit

mlx-community/Mistral-Large-4-1T-A52B-Text-4bit 是 ModelScope 魔搭社区上的 164.0B 参数text-generation模型，采用 other 许可，基于 mistralai/Mistral-Large-4-1T-A52B 构建。

- **Repository**: mlx-community/Mistral-Large-4-1T-A52B-Text-4bit
- **License**: other
- **Tasks**: text-generation
- **Parameters**: 164.0B
- **Base model**: mistralai/Mistral-Large-4-1T-A52B
- **Tags**: mlx
- **Downloads**: 5
- **Stars**: 1
- **Last updated**: 2026-10-29

Source: https://www.modelscope.cn/models/mlx-community/Mistral-Large-4-1T-A52B-Text-4bit

---

# mlx-community/Mistral-Large-4-1T-A52B-Text-4bit

This model [mlx-community/Mistral-Large-4-1T-A52B-Text-4bit](https://huggingface.co/mlx-community/Mistral-Large-4-1T-A52B-Text-4bit) was converted to MLX format from [mistralai/Mistral-Large-4-1T-A52B](https://huggingface.co/mistralai/Mistral-Large-4-1T-A52B) using mlx-lm with [ml-explore/mlx-lm#2018](https://github.com/ml-explore/mlx-lm/pull/2018). Refer to the [original model card](https://huggingface.co/mistralai/Mistral-Large-4-1T-A52B) for more details on the model.

This is a text-only conversion. The vision encoder and the DSpark draft head are not included.

The FP8 weights were dequantized and then quantized to 4-bit affine with group size 64 (4.5 bits per weight). The router and two indexer projections stay in bf16, as in the original checkpoint. It does not fit in the memory of one 512GB Mac. Peak memory was about 300 GB on each of two Macs in a short generation test with pipeline parallelism.

Tested with mlx-lm 0.32.3.dev13 and MLX 0.32.4.

## Use with mlx

```bash
pip install -U mlx-lm
```

Run it with pipeline parallelism over two Macs, for example with the [`sharded_generate.py`](https://github.com/ml-explore/mlx-lm/blob/main/examples/sharded_generate.py) example of mlx-lm:

```bash
mlx.launch --backend ring --hostfile hosts.json \
    sharded_generate.py --model mlx-community/Mistral-Large-4-1T-A52B-Text-4bit --pipeline --prompt "hello"
```

See the [MLX distributed guide](https://ml-explore.github.io/mlx/build/html/usage/distributed.html) for the host file and the network setup.