FaithJudge: leaderboard
The model answers from given passages only (summaries, question answering, business overviews from JSON), and o3-mini (high), prompted with human hallucination annotations for the same source, flags unsupported claims; the number is the share of the model's 511 responses flagged. Unlike Vectara's HHEM leaderboard, the judge is an LLM given annotated examples rather than a fine-tuned classifier.
Metric: Hallucination rate (%), pooled over 511 RAG responses (FaithBench and RAGTruth), judged by o3-mini (high) with human-annotated examples. 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 Flash | 6.26 | #467 | |
| 2 | Gemini 2.5 Pro | 6.65 | #274 | |
| 3 | Gemini 2.5 Pro (Preview 03-25) | 7.63 | #239 | |
| 4 | DeepSeek R1 0528 | 9.78 | #431 | |
| 5 | Gemini 2.0 Flash (001) | 10.18 | #727 | |
| 6 | GPT-4.1 | 11.94 | #505 | |
| 7 | GPT-4.5 (Preview) | 11.94 | #499 | |
| 8 | Claude Opus 4 (20250514) | Thinking | 14.09 | #300 |
| 9 | GPT-3.5 Turbo (0125) | 14.87 | #3575 | |
| 10 | Grok 3 | 15.26 | #606 | |
| 11 | GPT-4.1 Mini | 15.66 | #793 | |
| 12 | GPT-4o (2024-11-20) | 15.85 | #864 | |
| 13 | Claude 3.7 Sonnet (20250219) | 16.05 | #421 | |
| 14 | Llama 3.3 70B Instruct | 16.44 | #1387 | |
| 15 | Phi-4 | 17.03 | #1925 |
No result here: #1 Claude Fable 5.2, #2 GPT-6.2 Astra, #3 Gemini 4 Carbon.
Interactive version: theaggregate.ai/benchmark?slug=faithjudge · How It Works · Data refreshed daily, snapshot 2026-11-17.