LLM Hallucination Detection - Non-RAG Leaderboard — LLM Benchmark Hub

LLM Hallucination Detection - Non-RAG — leaderboard

Metric: Hallucination rate (%). Source: huggingface.co. 15 models tracked.

Top models

#	Model	Score
1	Magistral Small	28.07
2	Mistral Nemo Instruct (2407)	8.74
3	Mistral Small 3	7.85
4	Qwen 2.5 VL 7B Instruct	4.55
5	Kimi K2	3.04
6	Llama 4 Scout Instruct	2.48
7	Qwen 3 235B A22B FP8	2.09
8	DeepSeek V3 (0324)	0.91
9	Qwen 3 235B A22B 2507	0.82
10	DeepSeek R1 0528	0.78
11	Llama 4 Maverick Instruct FP8	0.69
12	GPT-4o	0.64
13	Claude Sonnet 4	0.6
14	Gemma 3 27B (IT)	0.48
15	Gemini 2.5 Pro	0.36

Interactive version: aibenchmarks.dev/benchmark?slug=llm-hallucination-detection-non-rag · How the rankings work · Data refreshed daily, snapshot 2026-07-08.