--- title: MAI-UI-8B-6bit canonical_url: "https://www.modelscope.cn/models/mlx-community/MAI-UI-8B-6bit" md_url: "https://www.modelscope.cn/models/mlx-community/MAI-UI-8B-6bit.md" repository: mlx-community/MAI-UI-8B-6bit last_updated: 2026-03-20 license: apache-2.0 pipeline_tag: image-text-to-text tasks: - image-text-to-text model_type: - qwen3_vl architectures: - Qwen3VLForConditionalGeneration base_model: - Tongyi-MAI/MAI-UI-8B base_model_relation: quantized parameters: 2.4B tensor_type: - F32 - U32 library_name: - mlx - safetensors - pytorch frameworks: - pytorch language: - en downloads: 15 stars: 0 tags: - mlx - mlx-vlm - safetensors - apple-silicon - conversational - gui - vision-language-model - qwen3_vl - mai-ui - grounding - mobile-navigation - computer-use - agent - 6-bit - quantized --- # MAI-UI-8B-6bit > MAI-UI-8B-6bit - mlx-community 在 ModelScope 开源的模型。This is a 6-bit quantized MLX conversion of Tongyi-MAI/MAI-UI-8B, optimized for Apple Silicon. mlx-community/MAI-UI-8B-6bit 是 ModelScope 魔搭社区上的 2.4B 参数image-text-to-text模型，采用 apache-2.0 许可，基于 Tongyi-MAI/MAI-UI-8B 构建。 - **Repository**: mlx-community/MAI-UI-8B-6bit - **License**: apache-2.0 - **Tasks**: image-text-to-text - **Parameters**: 2.4B - **Base model**: Tongyi-MAI/MAI-UI-8B - **Tags**: mlx, mlx-vlm, safetensors, apple-silicon, conversational, gui, vision-language-model, qwen3_vl, mai-ui, grounding, mobile-navigation, computer-use, agent, 6-bit, quantized - **Downloads**: 15 - **Stars**: 0 - **Last updated**: 2026-03-20 Source: https://www.modelscope.cn/models/mlx-community/MAI-UI-8B-6bit --- # MAI-UI-8B 6bit This is a 6-bit quantized MLX conversion of [Tongyi-MAI/MAI-UI-8B](https://huggingface.co/Tongyi-MAI/MAI-UI-8B), optimized for Apple Silicon. MAI-UI is a family of real-world centric foundation GUI agents built for grounding, GUI navigation, user interaction, and broader device-cloud agent workflows. The family spans multiple scales and is framed upstream around realistic deployment, including user interaction, MCP-style tool use, online RL, and device-cloud collaboration. This artifact was derived from the validated local MLX `bf16` reference conversion and then quantized with `mlx-vlm`. It was validated locally with both `mlx_vlm` prompt-packet checks and `vllm-mlx` OpenAI-compatible serve checks. ## Conversion Details | Field | Value | |---|---| | Upstream model | `Tongyi-MAI/MAI-UI-8B` | | Artifact type | `6bit quantized MLX conversion` | | Source artifact | local validated `bf16` MLX artifact | | Conversion tool | `mlx_vlm.convert` via `mlx-vlm 0.3.12` | | Python | `3.11.14` | | MLX | `0.31.0` | | Transformers | `5.2.0` | | Validation backend | `vllm-mlx (phase/p1 @ 8a5d41b)` | | Quantization | `6bit` | | Group size | `64` | | Quantization mode | `affine` | | Converter dtype note | `float32` | | Reported effective bits per weight | `8.644` | | Artifact size | `8.84G` | | Template repair | `tokenizer_config.json["chat_template"]` was re-injected from `chat_template.jinja` after quantization | Additional notes: - Root-level packaging is intentional for `vllm-mlx` multimodal detection compatibility. - `processor_config.json` and `video_preprocessor_config.json` are present at repo root. - This artifact intentionally augments tokenizer-visible template metadata for downstream compatibility checks. ## Validation This artifact passed local validation in this workspace: - `mlx_vlm` prompt-packet validation: `PASS` - `vllm-mlx` OpenAI-compatible serve validation: `PASS` Local validation notes: - output stayed in the same behavior envelope as the local `bf16` reference artifact - grounding shifted modestly relative to `bf16`, but still pointed at the correct `API Host` region - the known baseline schema limitation remained unchanged from `bf16`: the structured-action output still omitted the requested `reason` field ## Performance - Artifact size on disk: `8.84G` - Local fixed-packet `mlx_vlm` validation used about `34.87 GB` peak memory - Observed local fixed-packet throughput was about `150-166` prompt tok/s and `35.3-38.0` generation tok/s across the four validation prompts - Local `vllm-mlx` non-stream request time was about `27.83s`, slower than the `bf16` reference run These are local validation measurements, not a full benchmark suite. ## Usage ### Install ```bash pip install -U mlx-vlm ``` ### CLI ```bash python -m mlx_vlm.generate \ --model mlx-community/MAI-UI-8B-6bit \ --image path/to/image.png \ --prompt "Describe the visible controls on this screen." \ --max-tokens 256 \ --temperature 0.0 ``` ### Python ```python from mlx_vlm import load, generate model, processor = load("mlx-community/MAI-UI-8B-6bit") result = generate( model, processor, prompt="Describe the visible controls on this screen.", image="path/to/image.png", max_tokens=256, temp=0.0, ) print(result.text) ``` ### vllm-mlx Serve ```bash python -m vllm_mlx.cli serve mlx-community/MAI-UI-8B-6bit --mllm --localhost --port 8000 ``` ## Links - Upstream model: [Tongyi-MAI/MAI-UI-8B](https://huggingface.co/Tongyi-MAI/MAI-UI-8B) - Paper: [MAI-UI Technical Report: Real-World Centric Foundation GUI Agents](https://arxiv.org/abs/2512.22047) - Project page: [tongyi-mai.github.io/MAI-UI](https://tongyi-mai.github.io/MAI-UI/) - GitHub: [Tongyi-MAI/MAI-UI](https://github.com/Tongyi-MAI/MAI-UI) - MLX framework: [ml-explore/mlx](https://github.com/ml-explore/mlx) - mlx-vlm: [Blaizzy/mlx-vlm](https://github.com/Blaizzy/mlx-vlm) ## Other Quantizations Planned sibling repos in this wave: - [`mlx-community/MAI-UI-8B-bf16`](https://huggingface.co/mlx-community/MAI-UI-8B-bf16) - [`mlx-community/MAI-UI-8B-6bit`](https://huggingface.co/mlx-community/MAI-UI-8B-6bit) - this model - [`mlx-community/MAI-UI-8B-4bit`](https://huggingface.co/mlx-community/MAI-UI-8B-4bit) ## Notes and Limitations - This card reports local MLX conversion and validation results only. - Upstream benchmark claims belong to the original MAI-UI model family and were not re-run here unless explicitly stated. - Quantization changes numerical behavior relative to the local `bf16` reference artifact. - In local validation, the main change relative to `bf16` was latency rather than output collapse. ## Citation If you use this MLX conversion, please also cite the original MAI-UI work: ```bibtex @misc{zhou2025maiuitechnicalreportrealworld, title={MAI-UI Technical Report: Real-World Centric Foundation GUI Agents}, author={Hanzhang Zhou and Xu Zhang and Panrong Tong and Jianan Zhang and Liangyu Chen and Quyu Kong and Chenglin Cai and Chen Liu and Yue Wang and Jingren Zhou and Steven Hoi}, year={2025}, eprint={2512.22047}, archivePrefix={arXiv}, primaryClass={cs.CV}, url={https://arxiv.org/abs/2512.22047}, } ``` ## License This repo follows the upstream model license: Apache 2.0. See the upstream model card for the authoritative license details: [Tongyi-MAI/MAI-UI-8B](https://huggingface.co/Tongyi-MAI/MAI-UI-8B).