MobileGUI-VBench

MobileGUI-VBench is a comprehensive benchmark built by vivo AI Lab for evaluating mobile GUI agent capabilities. It is part of the BlueLM-GUI project — a real-device-centric flywheel for self-improving mobile GUI agents.

Overview

MobileGUI-VBench evaluates mobile GUI agents on 150 real-world tasks across 40 real-world mobile apps and 8 scenario categories:

Scenario	# Tasks
Video Streaming	24
Shopping & Deals	24
Lifestyle Services	23
Social & Communication	21
Travel & Transit	18
Music & Radio	17
Navigation	17
News & Reading	6

Key Features

• Task diversity: 140 single-app and 10 cross-app tasks.
• Instruction complexity: single-intent and multi-intent tasks with varying chain complexity (simple / medium / complex).
• Instruction types: explicit, implicit, and ambiguous instructions.
• Preconditions & clarifications: 65 tasks include environment preconditions, and 52 tasks require clarification-seeking behavior, testing the agent's ability to ask before acting.
• Function coverage: 77 distinct function points across 13 annotated dimensions.

Dataset Contents

File	Description
MobileGUI-VBench.jsonl	Task data in JSONL format (150 tasks)
MobileGUI-VBench.xlsx	Task data in Excel format
MobileGUI-VBench-Doc.html	Interactive documentation (English)
MobileGUI-VBench-说明文档.html	Interactive documentation (中文)
MobileGUI-VBench_table.md	Full benchmark comparison table
evaluation_framework.pdf	Evaluation framework report

Each task record contains the following fields:

{
  "task_id": 1,
  "query": "好久没上QQ了，看看朋友们最近都在发些什么动态",
  "scenario": "社交通讯",
  "involved_apps": "QQ",
  "precondition": "",
  "clarification_requirement": ""
}

• task_id: Unique task identifier
• query: Natural language instruction given to the agent
• scenario: One of the 8 scenario categories
• involved_apps: Apps required to complete the task (multiple apps separated by 、)
• precondition: Environment state required before the task can be executed
• clarification_requirement: Whether and when the agent should ask for clarification

Benchmark Results

Model	Params	MobileGUI-VBench	AndroidWorld	Access
BlueLM-GUI (Ours)	35B-A3B	87.4	84.9	In-house
Seed2.1-Pro	-	82.3	72.4	Closed
Seed2.0-Pro	-	80.8	71.5	Closed
Seed2.0-Lite	-	77.5	71.6	Closed
Gemini3.7-Flash	-	78.3	80.1	Closed
Claude Opus 4.8	-	56.9	63.8	Closed
GLM5.3-Flash	320B-A18B	66.4	78.4	Open
Qwen3.8-Max	2.4T-A95B	64.2	85.3	Open
Qwen3.8-Flash	125B-A6B	59.1	84.5	Open
Qwen3.8-27B	27B	58.0	81.9	Open
Qwen3.7-Plus	397B-A17B	49.6	81.0	Closed
Qwen3.6-35B-A3B	35B-A3B	40.7	73.3	Open
Qwen3.6-27B	27B	64.4	70.3	Open
Qwen3.5-122B-A10B	122B-A10B	57.3	66.4	Open
Qwen3.5-35B-A3B	35B-A3B	42.2	71.1	Open
Qwen3.5-27B	27B	53.7	64.2	Open
GUI-Owl-1.5-32B-Think	32B	47.1	71.6	Open
UI-Venus-2.0-27B	27B	-	84.0	Closed
UI-Venus-2.0-9B	9B	49.6	80.2	Open
UI-Venus-1.5-30B-A3B	30B-A3B	24.1	77.6	Open
PhoneBuddy	4B	39.0	83.2	Open
UI-Tars-2.0	230B-A23B	-	73.3	Closed
MAI-UI-235B-A22B	235B-A22B	-	76.7	Closed
Step-GUI	8B	-	80.2	Closed

Access

• Hugging Face: vivo-ai/MobileGUI-VBench
• GitHub: BlueLM-GUI/MobileGUI-VBench

Citation

If you use MobileGUI-VBench in your research, please cite:

@article{bluelm-gui,
  title   = {BlueLM-GUI Technical Report: A Real-Device-Centric Flywheel for Self-Improving Mobile GUI Agents},
  author  = {vivo AI Lab},
  journal = {arXiv preprint arXiv:2609.12394},
  year    = {2026},
  url     = {https://arxiv.org/abs/2609.12394}
}

License

This dataset is released under the Apache 2.0 License.