HalluScore: leaderboard
Large language models (LLMs) have achieved remarkable progress in natural language generation, but remain susceptible to hallucination. In response to growing concerns about hallucinations, several benchmarks have been developed, primarily in English and Chinese. However, Arabic remains underrepresented, with limited benchmarks for LLMs hallucination due to scarce annotated resources and the langu
Metric: Faith% (self-reported). Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 | 0.73 | #98 |
| 2 | GPT-4o | 0.85 | #343 |
| 3 | Claude Opus 4 | 0.85 | #163 |
| 4 | Grok 4 | 0.97 | #181 |
| 5 | Grok 4 Fast (Reasoning) | 0.97 | #254 (Grok 4 Fast) |
| 6 | Claude Sonnet 4.5 | 1.21 | #142 |
| 7 | Llama 4 Maverick Instruct FP8 | 1.33 | #414 |
| 8 | O4 Mini | 1.33 | #180 |
| 9 | Qwen 3 Next 80B A3B Instruct | 1.81 | #368 |
| 10 | DeepSeek V3 | 2.54 | #321 |
| 11 | O3 | 3.51 | #124 |
| 12 | DeepSeek R1 | 3.87 | #251 |
| 13 | ALLaM-7B-Instruct-preview | 3.87 | #718 |
| 14 | Fanar-1-9B | 10.04 | #778 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=halluscore · How It Works · Data refreshed daily, snapshot 2026-10-20.