Back to Arena
Cradle
by BAAI-Agents
System Card
OrganizationBAAI-Agents
Released2024-03
Architectureagentic-workflow / Skill registry + memory/logging for GCC
DetailsGeneral Computer Control (GCC) agent — takes screenshots, outputs keyboard/mouse events. Modules for skill registry, action planner, self-reflection, memory, and logging. Runs on RDR2, Stardew Valley, Chrome, Outlook, etc.
Parameters—
Domainagent-memoryepisodic-session
Open SourceYes
PaperView Paper
WebsiteVisit
CodeRepository
general-computer-controlscreenshot-agentskill-registrypaper-repo
Capability Profile
Benchmark Scores
6 of 14 benchmarksLong-Context Retrieval0/5
RULER
no dataNIAH
no dataLooGLE
no dataLongBench
no data∞Bench
no dataMulti-Turn Recall2/2
Cross-Session Memory1/1
Multi-Hop QA2/3
Agent Task Memory1/1
Personalization0/1
PerLTQA
no dataFactuality / Grounding0/1
RAGAS
no dataSources:Cradle paper (arXiv:2403.03186); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)Cradle paper (arXiv:2403.03186); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)Cradle paper (arXiv:2403.03186); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)Cradle paper (arXiv:2403.03186); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Cradle paper (arXiv:2403.03186); evaluated on MemoryBank: Enhancing LLMs with Long-Term Memory (Sun Yat-sen University, 2305)Cradle paper (arXiv:2403.03186); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)