TailNLG (Long-Tail Entities): leaderboard
Metric: chrF++ (0-100) against the TailNLG references for triples about long-tail (rare) Wikidata entities, averaged over English, Spanish and Italian; zero-shot prompt in the target language, 3 samples at temperature 0.7; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemma 3 12B (IT) | 60.38 | #655 |
| 2 | Gemma 3 4B (IT) | 54.32 | #971 |
| 3 | Qwen 2.5 7B Instruct | 54.27 | #846 |
| 4 | Llama 3.1 8B Instruct | 53.81 | #1018 |
| 5 | Qwen 2.5 3B Instruct | 50.1 | #1138 |
| 6 | Llama 3.2 3B Instruct | 49.75 | #1321 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=tailnlg-long-tail-entities · How It Works · Data refreshed daily, snapshot 2026-10-11.