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 MLX artifact was converted with mlx-vlm and 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-2B`
Artifact type	`bf16 MLX conversion`
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 @ 48b51ed)`
Quantization	`bf16`
Group size	`n/a`
Quantization mode	`n/a`
Artifact size	`4.0G`
Template repair	`tokenizer_config.json["chat_template"]` was re-injected from `chat_template.jinja` after conversion

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 refreshed artifact is intended to supersede the older nested mlx-community/MAI-UI-2B-bf16 line for clean serving-path and benchmark use.

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:

the Track A packet remained usable on the 2B line
grounding and structured-action rows stayed on the correct API Host target
the main remaining defect was action wording precision, not target collapse

Performance

Artifact size on disk: 4.0G
Local fixed-packet mlx_vlm validation used about 5.29 GB average peak memory
Observed local fixed-packet throughput was about 469-608 prompt tok/s and 2.5-15.2 generation tok/s across the four validation prompts

These are local validation measurements, not a full benchmark suite.

Usage

Install

pip install -U mlx-vlm

CLI

bash

python -m mlx_vlm.generate \
  --model mlx-community/MAI-UI-2B-bf16-v2 \
  --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-2B-bf16-v2")
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

python -m vllm_mlx.cli serve mlx-community/MAI-UI-2B-bf16-v2 --mllm --localhost --port 8000

Other Quantizations

Planned sibling repos in this Track C refresh:

mlx-community/MAI-UI-2B-bf16-v2 - this model
mlx-community/MAI-UI-2B-6bit-v2

Not published from this wave:

4bit was evaluated locally and rejected after runtime validation

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.
This refreshed bf16 artifact is the local reference artifact for the accepted 6bit variant in Track C.
Final public Track C comparative benchmarking happens after the refreshed 2B repos are uploaded.

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-2B.

mlx-community/MAI-UI-2B-bf16-v2

MAI-UI-2B bf16