FastGPT

by labring

System Card

Organizationlabring

Released2023-03

Architectureagentic-workflow / Knowledge-base LLM platform with workflow memory

DetailsVisual workflow designer for knowledge-base chat apps; hybrid search + reranking; MCP bi-directional support; iframe embedding. Conversation history is stored per app.

Parameters—

Domainrag-retrievalagent-memory

Open SourceYes

WebsiteVisit

CodeRepository

knowledge-basehybrid-searchmcplabring

Capability Profile

Benchmark Scores

6 of 14 benchmarks

Long-Context Retrieval

1/5

RULER

no data

NIAH

no data

LooGLE

no data

LongBench

603p

∞Bench

no data

Multi-Turn Recall

1/2

LoCoMo

80.187p

MemoryBank

no data

Cross-Session Memory

1/1

LongMemEval

77.460p

Multi-Hop QA

2/3

BABILong

no data

MultiHop-RAG

69.233p

HotpotQA

74.365p

Agent Task Memory

1/1

AgentBench-Mem

7226p

Personalization

0/1

PerLTQA

no data

Factuality / Grounding

0/1

RAGAS

no data

Sources:FastGPT (labring/FastGPT); evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)FastGPT (labring/FastGPT); evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)FastGPT (labring/FastGPT); evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)FastGPT (labring/FastGPT); evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)FastGPT (labring/FastGPT); evaluated on LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (Tsinghua KEG, 2308)FastGPT (labring/FastGPT); evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)