Kluster Hallucination Detection - RAG Resistance: leaderboard
The "RAG Resistance" subtask of Kluster Hallucination Detection. A task from the Kluster Hallucination Detection benchmark, measuring how reliably a model detects unsupported or fabricated claims.
Metric: Resistance (100 - Hallucination Rate %). Source: huggingface.co. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 98.43 | #152 |
| 2 | Claude Sonnet 4 | 97.79 | #199 |
| 3 | DeepSeek R1 0528 | 97.74 | #222 |
| 4 | Kimi K2 | 97.1 | #243 |
| 5 | Llama 4 Maverick Instruct FP8 | 96.66 | #414 |
| 6 | Gemma 3 27B (IT) | 96.29 | #523 |
| 7 | Llama 4 Scout Instruct | 95.77 | #539 |
| 8 | DeepSeek V3 (0324) | 95.34 | #340 |
| 9 | Mistral Small 3 | 95.26 | #627 |
| 10 | GPT-4o | 93.95 | #343 |
| 11 | Qwen 3 235B A22B FP8 | 93.84 | #301 |
| 12 | Qwen 3 235B A22B 2507 | 92.47 | #260 |
| 13 | Magistral Small | 91.38 | #1016 |
| 14 | Qwen 2.5 VL 7B Instruct | 90.65 | #654 |
| 15 | Mistral Nemo Instruct (2407) | 89.37 | #936 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=kluster-hallucination-detection-rag-resistance · How It Works · Data refreshed daily, snapshot 2026-10-20.