RAGragretrievalgroundednessevaluation
RAGBench
Benchmark suite for retrieval-augmented generation systems — evaluating retrieval quality, groundedness, and answer correctness on realistic RAG workloads.
Last reviewed: 23 July 2026
What it measures
End-to-end RAG quality: whether retrieved context is relevant and whether generated answers are faithful and correct.
Input / Output
Input: user query + corpus. Output: retrieved docs + generated answer. Metrics include retrieval recall, faithfulness, answer correctness.
Evaluation methodology
Separate retrieval and generation evaluation where possible. Use automated judges and/or gold annotations. Compare against naive RAG baselines.
Metadata
- Task
- RAG evaluation
- Domain
- Retrieval-augmented QA
- Modality
- Text
- Input type
- Query + document corpus
- Output type
- Retrieved passages + answer
- Evaluation type
- Retrieval + generation metrics
- Primary metric
- Answer correctness / faithfulness
- Secondary metrics
- Recall@k, Context precision, Groundedness
- Paper
- RAGBench: Explainable Benchmark for Retrieval-Augmented Generation Systems
- GitHub
- rungalileo/ragbench
- Dataset
- RAGBench (Galileo)
- Leaderboard
- Official live leaderboard (external)
Live leaderboard
Rankings change as new evaluations are published. View current results on the official leaderboard.
View live leaderboard →Datasets
GitHub repositories
Related research
Related guides
Related tools
Related models
Companies
Related rankings
Explore more benchmarks
- CRAGRAGComprehensive RAG benchmark from Meta stressing retrieval failures, outdated knowledge, and answer faithfulness across realistic industry scenarios.
- LongBenchLong ContextBilingual long-context benchmark covering single/multi-document QA, summarization, few-shot learning, synthetic tasks, and code across long inputs.
- HELMHolisticHolistic Evaluation of Language Models (Stanford CRFM): multi-metric, multi-scenario evaluation emphasizing transparency and taxonomy of use cases.
- GAIAAgentsGeneral AI Assistants benchmark: real-world questions requiring tool use, multi-step reasoning, and web/file interaction — hard for both humans and agents.
- Arena HardChatChallenging Chatbot Arena prompts packaged for automatic evaluation — a harder preference-style benchmark derived from real user battles.
- GPQAScienceGraduate-Level Google-Proof Q&A: PhD-level science questions designed so that non-experts with web search still struggle, testing deep domain reasoning.