reasoning benchmark
CRAG
CRAG (Comprehensive RAG Benchmark) is a factual question answering benchmark consisting of 4,409 question-answer pairs across 5 domains (finance, sports, music, movie, open domain) and 8 question categories. The benchmark includes mock APIs to simulate web and Knowledge Graph search, designed to represent the diverse and dynamic nature of real-world QA tasks with temporal dynamism ranging from years to seconds. It evaluates retrieval-augmented generation systems for trustworthy question answering.
Updated Aug 11, 2026
Models3
Model coverage3
MetricScore
EvidenceB
CRAG Ranking
Higher score ranks better on this benchmark.
| 01 | AM Nova ProAmazon | 50.3% | 100.0% | 3 | C | |
| 02 | AMNova LiteAmazon | 43.8% | 50.0% | 3 | C | |
| 03 | AMNova MicroAmazon | 43.1% | 0.0% | 3 | C |
CRAG Score Distribution
A closer view of the leading scores on this benchmark.
CRAG Highlights
The leading models and scores on this benchmark.
What is CRAG?
What CRAG measures and how its scores work.
CRAG (Comprehensive RAG Benchmark) is a factual question answering benchmark consisting of 4,409 question-answer pairs across 5 domains (finance, sports, music, movie, open domain) and 8 question categories. The benchmark includes mock APIs to simulate web and Knowledge Graph search, designed to represent the diverse and dynamic nature of real-world QA tasks with temporal dynamism ranging from years to seconds. It evaluates retrieval-augmented generation systems for trustworthy question answering.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- CRAG
- Modality
- text
- Primary category
- reasoning
- Score direction
- higher
- LLMBoard eligible
- Yes
- Evaluation key
- crag|llm-stats-current
Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about CRAG.
Which model scores highest on CRAG?
Nova Pro is currently ranked first with 50.3%.
What does CRAG measure?
CRAG (Comprehensive RAG Benchmark) is a factual question answering benchmark consisting of 4,409 question-answer pairs across 5 domains (finance, sports, music, movie, open domain) and 8 question categories. The benchmark includes mock APIs to simulate web and Knowledge Graph search, designed to represent the diverse and dynamic nature of real-world QA tasks with temporal dynamism ranging from years to seconds. It evaluates retrieval-augmented generation systems for trustworthy question answering.
Is a higher score better?
Yes. Higher values rank better for this benchmark.
How many models are compared?
3 model results are currently shown.
Does this benchmark affect the overall score?
Yes. This benchmark can contribute to the current LLMBoard capability score.