RHELM (RAG Top-20) - Fact: leaderboard
Metric: Accuracy (%; 207 fact questions, LLM judge, dialogue history plus external sources, dense top-20 retrieval). Source: arxiv.org. Saturation forecast: Around 2028. 4 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 | 50.2 |
| 2 | Claude Opus 4.5 | 50.2 |
| 3 | Gemini 2.5 Pro | 43 |
| 4 | GPT-4.1 Mini | 42.5 |
Interactive version: theaggregate.ai/benchmark?slug=rhelm-rag-top-20-fact · How It Works · Data refreshed daily, snapshot 2026-09-25.