Hallucinations Leaderboard: leaderboard
Public leaderboard evaluating LLM factuality, faithfulness, hallucination detection, instruction following, QA, reading comprehension, and summarization tasks.
Metric: Average Task Score (score). Source: huggingface.co. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 42 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | zephyr-7B-alpha | 52.08 | #1312 |
| 2 | Mistral 7B Instruct (v0.2) | 51.63 | #1333 |
| 3 | Mistral Instruct v0.1 7B | 50.48 | #1454 |
| 4 | Llama 2 13B Base | 49.78 | #1445 |
| 5 | zephyr-7B-beta | 49.62 | #1330 |
| 6 | Nous-Hermes-llama-2-7B | 46.65 | #1481 |
| 7 | Llama 2 7B Chat | 46.54 | #1499 |
| 8 | Yarn-Mistral-7B-128k | 46.29 | #1398 |
| 9 | Llama 2 13B Chat Base | 45.52 | #1422 |
| 10 | gemma-7B | 44.42 | #1384 |
| 11 | Orca-2-13B | 44.31 | #1448 |
| 12 | Llama 2 7B Base | 44.23 | #1518 |
| 13 | Mistral-7B-v0.1 | 44.06 | #1326 |
| 14 | SOLAR-10.7B-Instruct-v1.0 | 43.71 | #1075 |
| 15 | falcon-7B Instruct | 40.57 | #1622 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=hallucinations-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.