FaithJudge - RAGTruth Question Answering: leaderboard
Share of the model's 139 passage-grounded answers (RAGTruth QA) that the FaithJudge judge flags as hallucinated.
Metric: Hallucination rate (%) on 139 RAGTruth QA items. Source: raw.githubusercontent.com. Saturation forecast: Estimated already saturated. 35 models tracked. Updated 17 Nov 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 2.5 Pro (Preview 03-25) | 0.72 | #239 | |
| 2 | Gemini 2.0 Flash (001) | 0.72 | #727 | |
| 3 | Gemini 2.5 Pro | 1.44 | #274 | |
| 4 | Gemini 2.5 Flash | 2.88 | #467 | |
| 5 | GPT-4.1 Mini | 3.6 | #793 | |
| 6 | Llama 3.3 70B Instruct | 4.32 | #1387 | |
| 7 | GPT-4.1 | 4.32 | #505 | |
| 8 | Phi-4 | 4.32 | #1925 | |
| 9 | DeepSeek R1 0528 | 4.32 | #431 | |
| 10 | GPT-4o (2024-11-20) | 5.04 | #864 | |
| 11 | Claude Sonnet 4 (20250514) | 5.04 | #405 | |
| 12 | Phi-4 Mini Instruct | 5.04 | #6499 | |
| 13 | GPT-4.5 (Preview) | 5.04 | #499 | |
| 14 | Claude Opus 4 (20250514) | Thinking | 5.04 | #300 |
| 15 | GPT-3.5 Turbo (0125) | 5.76 | #3575 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=faithjudge-ragtruth-question-answering · How It Works · Data refreshed daily, snapshot 2026-11-17.