OSWorld provides the first unified real-computer benchmark for open-ended multimodal agent tasks, exposing large performance gaps between humans and state-of-the-art LLM/VLM agents.
hub Canonical reference
Mobile-Agent: Autonomous Multi-Modal Mobile Device Agent with Visual Perception
Canonical reference. 91% of citing Pith papers cite this work as background.
29
Pith papers citing it
Background
91% of classified citations
abstract
Mobile device agent based on Multimodal Large Language Models (MLLM) is becoming a popular application. In this paper, we introduce Mobile-Agent, an autonomous multi-modal mobile device agent. Mobile-Agent first leverages visual perception tools to accurately identify and locate both the visual and textual elements within the app's front-end interface. Based on the perceived vision context, it then autonomously plans and decomposes the complex operation task, and navigates the mobile Apps through operations step by step. Different from previous solutions that rely on XML files of Apps or mobile system metadata, Mobile-Agent allows for greater adaptability across diverse mobile operating environments in a vision-centric way, thereby eliminating the necessity for system-specific customizations. To assess the performance of Mobile-Agent, we introduced Mobile-Eval, a benchmark for evaluating mobile device operations. Based on Mobile-Eval, we conducted a comprehensive evaluation of Mobile-Agent. The experimental results indicate that Mobile-Agent achieved remarkable accuracy and completion rates. Even with challenging instructions, such as multi-app operations, Mobile-Agent can still complete the requirements. Code and model will be open-sourced at https://github.com/X-PLUG/MobileAgent.
hub tools
citation-role summary
background 10
baseline 1
citation-polarity summary
representative citing papers
WinDeskGround is a parametrically generated benchmark of 1,356 instruction-target pairs that reveals accuracy declines in state-of-the-art MLLMs under partial occlusion in multi-window GUI settings.
VLMs detect primitive motion in UI animations reliably but show inconsistent high-level interpretation of purposes and meanings, with large gaps relative to human performance.
AgenTEE isolates LLM agent runtime, inference, and apps in independently attested cVMs on Arm-based edge devices, achieving under 5.15% overhead versus commodity OS deployments.
The work creates a new benchmark for humanizing GUI agent touch dynamics via a MinMax detector-agent model, a mobile touch dataset, and methods showing agents can match human behavior without losing task performance.
AQuaUI uses adaptive quadtrees to cut visual tokens in GUI-agent LMMs by up to 29.52% at inference time while retaining 99.06% of full-token accuracy on grounding and navigation benchmarks.
ICRL uses joint RL training of solver and critic with distribution-calibration re-weighting and role-wise advantage estimation to internalize critique into unassisted LLM performance, yielding 6.4-point gains on agentic tasks and 7.0 on math reasoning with Qwen3 models.
CAVI framework uses character-guided token pruning, orthogonal feature modulation, and modality-adaptive role steering to resolve modality-role interference in multimodal RPAs.
VLAA-GUI adds mandatory visual verifiers, multi-tier loop breakers, and on-demand search to GUI agents, reaching 77.5% on OSWorld and 61.0% on WindowsAgentArena with some models exceeding human performance.
A new benchmark shows LLM smartphone agents achieve comparable success with screen text alone as with screenshots, but both fail often due to UI accessibility and reasoning gaps.
SkillDroid compiles LLM-guided GUI trajectories into parameterized skill templates and replays them via a matching cascade, reaching 85.3% success rate with 49% fewer LLM calls and improving from 87% to 91% over 150 rounds while the stateless baseline drops to 44%.
EdgeFlow reduces mobile LLM cold-start latency up to 4.07x versus llama.cpp, MNN, and llm.npu by NPU-aware adaptive quantization, SIMD-friendly packing, and synergistic granular CPU-NPU pipelining at comparable accuracy.
AgentProg reframes interaction history as a program with variables and control flow, plus a belief state for partial observability, achieving SOTA success rates on long-horizon GUI benchmarks while baselines degrade.
UMDAM introduces a column-major tile-based data layout and configurable DRAM mapping to enable efficient NPU-PIM co-execution for LLM inference, reducing TTFT by up to 3.0x and TTLT by 2.18x on OPT models without added memory overhead or bandwidth loss.
DroidRetriever is a transparent steerable mobile automation system that decomposes information-seeking tasks with multi-LLM agents, navigates apps, synthesizes reports with screenshots, and provides a dashboard for real-time user intervention and privacy pauses.
Personal agents require edge deployment to preserve high-fidelity local context and zero-latency loops, as claimed through three structural shifts away from cloud-centric designs.
SE-GA combines Test-Time Memory Extension for dynamic context retrieval with Memory-Augmented Self-Evolution training to reach 89.0% on ScreenSpot and 75.8% on AndroidControl-High.
DRS-GUI introduces a dynamic region search method with Focus/Shift/Scatter actions and MCTS-based planning that improves GUI grounding accuracy by 14% on ScreenSpot-Pro for both general and GUI-specific MLLMs without any training.
Degradation-Driven Prompting improves VQA by intentionally reducing image detail and using masks, lines, and examples to guide models toward essential structures.
VisionClaw couples continuous egocentric vision on smart glasses with speech-driven AI agents to enable hands-free real-world tasks, with lab and field studies showing faster completion and a shift toward opportunistic delegation.
Empirical study finds background semantics, random pruning, and recency-based allocation improve token efficiency for GUI visual agents.
TwigVLM adds a twig module to VLMs for twig-guided token pruning and self-speculative decoding, retaining 96% performance after pruning 88.9% visual tokens and delivering 154% speedup on long responses for LLaVA-1.5-7B.
A survey of 87 agents for computer use and 33 datasets that introduces a three-dimensional taxonomy across domain, interaction, and agent perspectives and identifies six research gaps.
World models trained on delta text, full text, diffusion images, and renderable code achieve SoTA on two benchmarks and improve downstream GUI agent performance on three mobile datasets with modality-specific strengths.
citing papers explorer
-
OSWorld: Benchmarking Multimodal Agents for Open-Ended Tasks in Real Computer Environments
OSWorld provides the first unified real-computer benchmark for open-ended multimodal agent tasks, exposing large performance gaps between humans and state-of-the-art LLM/VLM agents.
-
WinDeskGround: A Benchmark for Robust GUI Grounding in Complex Multi-Window Desktop Environments
WinDeskGround is a parametrically generated benchmark of 1,356 instruction-target pairs that reveals accuracy declines in state-of-the-art MLLMs under partial occlusion in multi-window GUI settings.
-
Beyond Screenshots: Evaluating VLMs' Understanding of UI Animations
VLMs detect primitive motion in UI animations reliably but show inconsistent high-level interpretation of purposes and meanings, with large gaps relative to human performance.
-
AgenTEE: Confidential LLM Agent Execution on Edge Devices
AgenTEE isolates LLM agent runtime, inference, and apps in independently attested cVMs on Arm-based edge devices, achieving under 5.15% overhead versus commodity OS deployments.
-
Turing Test on Screen: A Benchmark for Mobile GUI Agent Humanization
The work creates a new benchmark for humanizing GUI agent touch dynamics via a MinMax detector-agent model, a mobile touch dataset, and methods showing agents can match human behavior without losing task performance.
-
AQuaUI: Visual Token Reduction for GUI Agents with Adaptive Quadtrees
AQuaUI uses adaptive quadtrees to cut visual tokens in GUI-agent LMMs by up to 29.52% at inference time while retaining 99.06% of full-token accuracy on grounding and navigation benchmarks.
-
ICRL: Learning to Internalize Self-Critique with Reinforcement Learning
ICRL uses joint RL training of solver and critic with distribution-calibration re-weighting and role-wise advantage estimation to internalize critique into unassisted LLM performance, yielding 6.4-point gains on agentic tasks and 7.0 on math reasoning with Qwen3 models.
-
Through the Lens of Character: Resolving Modality-Role Interference in Multimodal Role-Playing Agent
CAVI framework uses character-guided token pruning, orthogonal feature modulation, and modality-adaptive role steering to resolve modality-role interference in multimodal RPAs.
-
VLAA-GUI: Knowing When to Stop, Recover, and Search, A Modular Framework for GUI Automation
VLAA-GUI adds mandatory visual verifiers, multi-tier loop breakers, and on-demand search to GUI agents, reaching 77.5% on OSWorld and 61.0% on WindowsAgentArena with some models exceeding human performance.
-
Do LLMs Need to See Everything? A Benchmark and Study of Failures in LLM-driven Smartphone Automation using Screentext vs. Screenshots
A new benchmark shows LLM smartphone agents achieve comparable success with screen text alone as with screenshots, but both fail often due to UI accessibility and reasoning gaps.
-
SkillDroid: Compile Once, Reuse Forever
SkillDroid compiles LLM-guided GUI trajectories into parameterized skill templates and replays them via a matching cascade, reaching 85.3% success rate with 49% fewer LLM calls and improving from 87% to 91% over 150 rounds while the stateless baseline drops to 44%.
-
EdgeFlow: Fast Cold Starts for LLMs on Mobile Devices
EdgeFlow reduces mobile LLM cold-start latency up to 4.07x versus llama.cpp, MNN, and llm.npu by NPU-aware adaptive quantization, SIMD-friendly packing, and synergistic granular CPU-NPU pipelining at comparable accuracy.
-
AgentProg: Empowering Long-Horizon GUI Agents with Program-Guided Context Management
AgentProg reframes interaction history as a program with variables and control flow, plus a belief state for partial observability, achieving SOTA success rates on long-horizon GUI benchmarks while baselines degrade.
-
UMDAM: A Unified Data Layout and DRAM Address Mapping for Heterogenous NPU-PIM
UMDAM introduces a column-major tile-based data layout and configurable DRAM mapping to enable efficient NPU-PIM co-execution for LLM inference, reducing TTFT by up to 3.0x and TTLT by 2.18x on OPT models without added memory overhead or bandwidth loss.
-
DroidRetriever: A Transparent and Steerable Automation System for Collaborative Mobile Information Seeking
DroidRetriever is a transparent steerable mobile automation system that decomposes information-seeking tasks with multi-LLM agents, navigates apps, synthesizes reports with screenshots, and provides a dashboard for real-time user intervention and privacy pauses.
-
Beyond Scaling: Agents Are Heading to the Edge
Personal agents require edge deployment to preserve high-fidelity local context and zero-latency loops, as claimed through three structural shifts away from cloud-centric designs.
-
SE-GA: Memory-Augmented Self-Evolution for GUI Agents
SE-GA combines Test-Time Memory Extension for dynamic context retrieval with Memory-Augmented Self-Evolution training to reach 89.0% on ScreenSpot and 75.8% on AndroidControl-High.
-
DRS-GUI: Dynamic Region Search for Training-Free GUI Grounding
DRS-GUI introduces a dynamic region search method with Focus/Shift/Scatter actions and MCTS-based planning that improves GUI grounding accuracy by 14% on ScreenSpot-Pro for both general and GUI-specific MLLMs without any training.
-
Less Detail, Better Answers: Degradation-Driven Prompting for VQA
Degradation-Driven Prompting improves VQA by intentionally reducing image detail and using masks, lines, and examples to guide models toward essential structures.
-
VisionClaw: Always-On AI Agents through Smart Glasses
VisionClaw couples continuous egocentric vision on smart glasses with speech-driven AI agents to enable hands-free real-world tasks, with lab and field studies showing faster completion and a shift toward opportunistic delegation.
-
Rethinking Token Pruning for Historical Screenshots in GUI Visual Agents: Semantic, Spatial, and Temporal Perspectives
Empirical study finds background semantics, random pruning, and recency-based allocation improve token efficiency for GUI visual agents.
-
Growing a Multi-head Twig via Distillation and Reinforcement Learning to Accelerate Large Vision-Language Models
TwigVLM adds a twig module to VLMs for twig-guided token pruning and self-speculative decoding, retaining 96% performance after pruning 88.9% visual tokens and delivering 154% speedup on long responses for LLaVA-1.5-7B.
-
A Comprehensive Survey of Agents for Computer Use: Foundations, Challenges, and Future Directions
A survey of 87 agents for computer use and 33 datasets that introduces a three-dimensional taxonomy across domain, interaction, and agent perspectives and identifies six research gaps.
-
How Mobile World Model Guides GUI Agents?
World models trained on delta text, full text, diffusion images, and renderable code achieve SoTA on two benchmarks and improve downstream GUI agent performance on three mobile datasets with modality-specific strengths.
-
Securing Computer-Use Agents: A Unified Architecture-Lifecycle Framework for Deployment-Grounded Reliability
The paper develops a unified framework that organizes computer-use agent reliability around perception-decision-execution layers and creation-deployment-operation-maintenance stages to map security and alignment interventions.
-
ClawMobile: Rethinking Smartphone-Native Agentic Systems
ClawMobile proposes a hierarchical system separating probabilistic LLM planning from structured deterministic execution to improve stability and reproducibility of agentic systems on real smartphones.
-
Large Language Model-Brained GUI Agents: A Survey
A survey consolidating frameworks, data practices, large action models, benchmarks, applications, and research gaps in LLM-brained GUI agents.
-
X-OmniClaw Technical Report: A Unified Mobile Agent for Multimodal Understanding and Interaction
Describes X-OmniClaw, a multimodal mobile agent architecture using Omni Perception, Memory, and Action modules with behavior cloning for Android task execution.
-
A Survey on Multimodal Large Language Models
This survey organizes the architectures, training strategies, data, evaluation methods, extensions, and challenges of Multimodal Large Language Models.