---
license: apache-2.0
base_model: inclusionAI/VISTA-4B
base_model_relation: quantized
pipeline_tag: text-generation
library_name: mlx
tags:
- mlx
- qwen3_5
- text-generation
---

# VISTA-4B-MLX-4bit

MLX (Apple Silicon) conversion of [inclusionAI/VISTA-4B](https://huggingface.co/inclusionAI/VISTA-4B), quantized to **4-bit**. First MLX build of this model. Text-only build of the backbone.

## Quantizations
Part of the [**VISTA MLX** collection](https://huggingface.co/collections/pipenetwork/vista-mlx-6a31a7c3e3a16328417b6064).

| Variant | |
|---|---|
| [8-bit](https://huggingface.co/pipenetwork/VISTA-4B-MLX-8bit) | |
| [6-bit](https://huggingface.co/pipenetwork/VISTA-4B-MLX-6bit) | |
| [5-bit](https://huggingface.co/pipenetwork/VISTA-4B-MLX-5bit) | |
| **4-bit** (this repo) | |

## Use with mlx-lm
```bash
pip install mlx-lm
python -m mlx_lm generate --model pipenetwork/VISTA-4B-MLX-4bit --prompt "Hello" -m 200
```

## Validation
Smoke-tested locally: loads and generates coherent text.

## License
`apache-2.0` (inherited from base). Quantization config: `{"group_size": 64, "bits": 4, "mode": "affine"}`.