FaithJudge - RAGTruth Summarization: leaderboard

Share of the model's 150 RAGTruth news summaries that the FaithJudge judge flags as hallucinated.

Metric: Hallucination rate (%) on 150 RAGTruth summaries. Source: raw.githubusercontent.com. Saturation forecast: Estimated already saturated. 35 models tracked. Updated 17 Nov 2026.

Top models

#ModelSettingScoreOverall rank
1Gemini 2.5 Pro4.67#274
2Gemini 2.5 Flash4.67#467
3Gemini 2.0 Flash (001)6.67#727
4Phi-48#1925
5DeepSeek R1 05288#431
6Llama 3.3 70B Instruct8.67#1387
7GPT-3.5 Turbo (0125)8.67#3575
8Gemini 2.5 Pro (Preview 03-25)9.33#239
9glm-4-9B-chat-hf9.33#3133
10GPT-4o (2024-11-20)10#864
11Mistral Small 310#1799
12GPT-4.5 (Preview)10#499
13GPT-4.1 Mini10.67#793
14Claude 3.7 Sonnet (20250219)Thinking10.67#421
15GPT-4.111.33#505

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-summarization · How It Works · Data refreshed daily, snapshot 2026-11-17.