Back to Arena

Galileo AI

by Galileo Technologies Inc.

System Card

OrganizationGalileo Technologies Inc.
Released2021-06
Architectureagentic-workflow / LLM evaluation with Luna foundation model
DetailsGalileo's core differentiator is Luna, a family of compact Evaluation Foundation Models fine-tuned specifically for hallucination detection, toxicity, prompt security, and data leak detection. ChainPoll methodology achieves 85% correlation with human feedback. Luna replaces expensive LLM-as-judge calls with low-latency, low-cost specialized models. Also offers guardrails, data curation, and issue triage.
Parameters—
Domainrag-retrievalagent-memory
Open SourceNo
WebsiteVisit
Luna-EFMhallucination-detectionguardrailsChainPollobservability

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
76.866p
MemoryBank
no data
Cross-Session Memory
1/1
LongMemEval
78.970p
Multi-Hop QA
2/3
BABILong
no data
MultiHop-RAG
73.665p
HotpotQA
76.278p
Agent Task Memory
1/1
AgentBench-Mem
7226p
Personalization
0/1
PerLTQA
no data
Factuality / Grounding
0/1
RAGAS
no data
Sources:Galileo AI vendor documentation; evaluated on HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering (Stanford / CMU, 1809)Galileo AI vendor documentation; evaluated on MultiHop-RAG: Benchmarking Retrieval-Augmented Generation for Multi-Hop Queries (HKUST, 2401)Galileo AI vendor documentation; evaluated on AgentBench Memory Track (Tsinghua KEG, 2308)Galileo AI vendor documentation; evaluated on LoCoMo: Long-Term Conversational Memory Benchmark (Snap Research, 2402)Galileo AI vendor documentation; evaluated on LongBench: A Bilingual, Multitask Benchmark for Long Context Understanding (Tsinghua KEG, 2308)Galileo AI vendor documentation; evaluated on LongMemEval: Benchmarking Chat Assistants on Long-Term Interactive Memory (Salesforce AI Research, 2410)