---
title: UI-TARS-1.5-7B-8bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/UI-TARS-1.5-7B-8bit"
md_url: "https://www.modelscope.cn/models/mlx-community/UI-TARS-1.5-7B-8bit.md"
repository: mlx-community/UI-TARS-1.5-7B-8bit
last_updated: 2026-05-31
license: apache-2.0
pipeline_tag: image-text-to-text
tasks:
  - image-text-to-text
model_type:
  - qwen2_5_vl
architectures:
  - Qwen2_5_VLForConditionalGeneration
base_model:
  - ByteDance-Seed/UI-TARS-1.5-7B
base_model_relation: quantized
parameters: 2.8B
tensor_type:
  - U32
  - BF16
library_name:
  - mlx
  - safetensors
  - pytorch
frameworks:
  - pytorch
language:
  - en
downloads: 33
stars: 1
tags:
  - multimodal
  - gui
  - gui-agent
  - ui-tars
  - mlx
---

# UI-TARS-1.5-7B-8bit

> UI-TARS-1.5-7B-8bit - mlx-community 在 ModelScope 开源的模型。UI-TARS-1.5-7B-8bit (MLX)

mlx-community/UI-TARS-1.5-7B-8bit 是 ModelScope 魔搭社区上的 2.8B 参数image-text-to-text模型，采用 apache-2.0 许可，基于 ByteDance-Seed/UI-TARS-1.5-7B 构建。

- **Repository**: mlx-community/UI-TARS-1.5-7B-8bit
- **License**: apache-2.0
- **Tasks**: image-text-to-text
- **Parameters**: 2.8B
- **Base model**: ByteDance-Seed/UI-TARS-1.5-7B
- **Tags**: multimodal, gui, gui-agent, ui-tars, mlx
- **Downloads**: 33
- **Stars**: 1
- **Last updated**: 2026-05-31

Source: https://www.modelscope.cn/models/mlx-community/UI-TARS-1.5-7B-8bit

---

# UI-TARS-1.5-7B-8bit (MLX)

8-bit MLX quantization of [ByteDance-Seed/UI-TARS-1.5-7B](https://huggingface.co/ByteDance-Seed/UI-TARS-1.5-7B),
a Qwen2.5-VL vision-language GUI-agent model for grounded computer/browser control.

- **Quantization:** 8-bit affine, group size 64 (~9.11 effective bits/weight)
- **Architecture:** `Qwen2_5_VLForConditionalGeneration` (`qwen2_5_vl`) — ViT vision encoder + Qwen2.5 LM
- **Converted with:** `mlx_vlm.convert` (mlx-vlm)

## Use with mlx-vlm

```bash
pip install mlx-vlm
python -m mlx_vlm generate \
  --model mlx-community/UI-TARS-1.5-7B-8bit \
  --image screenshot.png \
  --prompt "Click the search box." \
  --max-tokens 128
```

This 8-bit build complements the existing
[`mlx-community/UI-TARS-1.5-7B-4bit`](https://huggingface.co/mlx-community/UI-TARS-1.5-7B-4bit)
and [`-6bit`](https://huggingface.co/mlx-community/UI-TARS-1.5-7B-6bit) — higher fidelity at ~8.8 GB.