Back to Arena

ChatDev 2.0

by OpenBMB

System Card

OrganizationOpenBMB
Released2026-01
Architectureagentic-workflow / Multi-agent workflow with memory module
DetailsZero-code multi-agent platform. Backend is FastAPI with a runtime manager; workflow engine runs YAML entity configs; frontend Vue 3. Persistent memory module tracks context flow, intermediate artifacts, and variable passing between agent nodes.
Parameters—
Domainagent-memoryepisodic-session
Open SourceYes
PaperView Paper
WebsiteVisit
CodeRepository
multi-agentzero-codeworkflowopenbmb

Capability Profile

Benchmark Scores

6 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
2/2
LoCoMo
76.160p
MemoryBank
72.327p
Cross-Session Memory
1/1
LongMemEval
79.272p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
70.541p
HotpotQA
71.852p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:ChatDev 2.0 paper (arXiv:2307.07924); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)ChatDev 2.0 paper (arXiv:2307.07924); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)ChatDev 2.0 paper (arXiv:2307.07924); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)ChatDev 2.0 paper (arXiv:2307.07924); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)ChatDev 2.0 paper (arXiv:2307.07924); evaluated on MemoryBank: Enhancing LLMs with Long-Term Memory (Sun Yat-sen University, 2305)ChatDev 2.0 paper (arXiv:2307.07924); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)