Halluverse-M3 — leaderboard
Multitask multilingual hallucination detection benchmark spanning QA and dialogue summarization in English, Arabic, Hindi, and Turkish.
Metric: Macro Accuracy (self-reported). Source: benchmarklist.com. Status: saturation imminent. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 80.3 |
| 2 | GPT-4.1 | 78.66 |
| 3 | Claude 3.5 Sonnet (20240620) | 76.98 |
| 4 | DeepSeek V2.5 | 73.76 |
| 5 | GPT-4o Mini | 73.39 |
| 6 | Llama 3.3 70B | 70.53 |
| 7 | Phi-4 | 70.38 |
| 8 | Qwen 2.5 72B Instruct | 69.81 |
| 9 | Qwen 2.5 32B | 66.84 |
| 10 | Gemma 2 27B | 62.1 |
| 11 | Qwen 2.5 7B | 59.4 |
| 12 | Mistral 7B | 56.06 |
Interactive version: theaggregate.ai/benchmark?slug=halluverse-m3 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.