Open PL LLM - CBD (generative, 5-shot): leaderboard
Cyberbullying detection in Polish tweets (PolEval 2019). Open PL LLM Leaderboard (SpeakLeash), 5-shot, generative variant: the answer is extracted from generated text; macro F1 (%), higher is better. Runs not available on the Hugging Face Hub are excluded.
Metric: Macro F1 (%). Source: huggingface.co. Saturation forecast: Around May 2028. 319 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Mistral Large 2 (Nov) Instruct (2411) | 43.98 | #480 |
| 2 | Llama 3.1 405B Instruct FP8 | 42.74 | #400 |
| 3 | Qwen 2.5 72B | 40.96 | #512 |
| 4 | Qwen 2 72B Instruct | 39.82 | #579 |
| 5 | Bielik-11B-v2.3-Instruct | 38.78 | #700 |
| 6 | Mistral Large 2 (Jul) | 38.7 | #490 |
| 7 | Llama 4 Scout Instruct | 38.46 | #547 |
| 8 | Mixtral 8x22B Instruct | 38.19 | #738 |
| 9 | Qwen 2.5 72B Instruct | 38.13 | #469 |
| 10 | Llama 3.1 70B Instruct | 38.11 | #567 |
| 11 | Llama 3.1 Nemotron 70B Instruct | 37.85 | #563 |
| 12 | aya-expanse-32B | 37.62 | #687 |
| 13 | Qwen 2.5 32B Instruct | 37.49 | #509 |
| 14 | QwQ 32B-Preview | 37.39 | #671 |
| 15 | Phi-4 | 37.27 | #707 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-pl-llm-cbd-generative-5-shot · How It Works · Data refreshed daily, snapshot 2026-11-05.