EnterpriseRAG Bench - Valid Extra Docs Resistance: leaderboard
The "Valid Extra Docs Resistance" subtask of EnterpriseRAG Bench. Onyx's enterprise retrieval-augmented generation benchmark — question answering over realistic company document corpora, scoring answer correctness, completeness, recall and resistance to plausible-but-irrelevant extra documents.
Metric: Resistance (100 - Invalid Extra Docs %). Source: huggingface.co. 34 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | CDL (Causal Dynamics Lab) | 99.76 | |
| 2 | Mixedbread (+ Opus 5) | 99.61 | |
| 3 | Nacre.io | 99.58 | |
| 4 | OpenClaw | 99.53 | |
| 5 | MemOnDemand | 99.51 | |
| 6 | Flexor.ai (+ GPT‑6.1 Sol) | 99.48 | |
| 7 | fgroo | 99.37 | |
| 8 | xFloor.ai | 99.27 | |
| 9 | Neurond AI (neurond.com) | 99.23 | |
| 10 | AHOY Labs | 98.97 |
Interactive version: theaggregate.ai/benchmark?slug=enterpriserag-bench-valid-extra-docs-resistance · How It Works · Data refreshed daily, snapshot 2026-10-20.