Back to Arena

Mobile-Agent

by Alibaba Tongyi Lab (X-PLUG)

System Card

OrganizationAlibaba Tongyi Lab (X-PLUG)
Released2024-01
Architectureagentic-workflow / Multi-platform GUI agent with long-horizon memory
DetailsFamily of GUI agents (GUI-Owl-1.5 models 2B-235B) for desktop, mobile, and browser automation. v3.5 emphasizes end-to-end task execution and long-horizon memory across cross-app workflows.
Parameters—
Domainagent-memory
Open SourceYes
PaperView Paper
CodeRepository
gui-agentgui-owllong-horizonalibaba

Capability Profile

Benchmark Scores

5 of 14 benchmarks
Long-Context Retrieval
0/5
RULER
no data
NIAH
no data
LooGLE
no data
LongBench
no data
∞Bench
no data
Multi-Turn Recall
1/2
LoCoMo
7439p
MemoryBank
no data
Cross-Session Memory
1/1
LongMemEval
80.181p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
69.434p
HotpotQA
77.486p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:Mobile-Agent paper (arXiv:2508.15144); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)Mobile-Agent paper (arXiv:2508.15144); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Mobile-Agent paper (arXiv:2508.15144); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)Mobile-Agent paper (arXiv:2508.15144); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)Mobile-Agent paper (arXiv:2508.15144); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)