---
title: UI-Venus-Navi-7B-4bit
canonical_url: "https://www.modelscope.cn/models/mlx-community/UI-Venus-Navi-7B-4bit"
md_url: "https://www.modelscope.cn/models/mlx-community/UI-Venus-Navi-7B-4bit.md"
repository: mlx-community/UI-Venus-Navi-7B-4bit
last_updated: 2025-09-21
license: apache-2.0
model_type:
  - qwen2_5_vl
architectures:
  - Qwen2_5_VLForConditionalGeneration
base_model:
  - inclusionAI/UI-Venus-Navi-7B
base_model_relation: quantized
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
inference_backends:
  - "deploy_task vlm/text/emb"
  - "lmdeploy 0.9.1"
  - "sglang 0.5.2"
  - "vllm 0.9.2"
downloads: 84
stars: 0
---

# UI-Venus-Navi-7B-4bit

> UI-Venus-Navi-7B-4bit - mlx-community 在 ModelScope 开源的模型。Model Description This model is covnert by mlxvlm from inclusionAI/UI-Venus-Navi-7B.

mlx-community/UI-Venus-Navi-7B-4bit 是 ModelScope 魔搭社区上的机器学习模型，采用 apache-2.0 许可，基于 inclusionAI/UI-Venus-Navi-7B 构建，可用 deploy_task vlm/text/emb、lmdeploy 0.9.1、sglang 0.5.2 部署。

- **Repository**: mlx-community/UI-Venus-Navi-7B-4bit
- **License**: apache-2.0
- **Base model**: inclusionAI/UI-Venus-Navi-7B
- **Inference backends**: deploy_task vlm/text/emb, lmdeploy 0.9.1, sglang 0.5.2, vllm 0.9.2
- **Downloads**: 84
- **Stars**: 0
- **Last updated**: 2025-09-21

Source: https://www.modelscope.cn/models/mlx-community/UI-Venus-Navi-7B-4bit

---

## Model Description
This model is covnert by mlx_vlm from [inclusionAI/UI-Venus-Navi-7B](https://huggingface.co/inclusionAI/UI-Venus-Navi-7B).

UI-Venus is a native UI agent based on the Qwen2.5-VL multimodal large language model, designed to perform precise GUI element grounding and effective navigation using only screenshots as input. It achieves state-of-the-art performance through Reinforcement Fine-Tuning (RFT) with high-quality training data.

 <p align="center">
  📈 UI-Venus Benchmark Performance
</p>

<p align="center">
  <img src="https://huggingface.co/inclusionAI/UI-Venus-Navi-7B/resolve/main/performance_venus.png" alt="UI-Venus Performance Across Datasets" width="1200" />
  <br>
</p>

The 7B model achieve 94.1%/50.8% on Screenspot-V2 and Screenspot-Pro benchmarks, surpassing prior SOTA models such as GTA1 and UI-TARS-1.5-7B. On the AndroidWorld navigation benchmark, it achieve 49.1% success rates, respectively, demonstrating strong planning and generalization capabilities

## Quick Start
```shell
mlx_vlm.generate --model mlx-community/UI-Venus-Navi-7B-4bit \
  --max-tokens 1024 \
  --temperature 0.0 \
  --prompt "List all contacts’ names and their corresponding grounding boxes([x1, y1, x2, y2]) from the left sidebar of the IM chat interface, return the results in JSON format." \
  --image https://wechat.qpic.cn/uploads/2016/05/WeChat-Windows-2.11.jpg
```