ThaiSafetyBench - Discrimination, Exclusion, Toxicity, Hateful, Offensive: leaderboard
Metric: Attack Success Rate (%). Source: huggingface.co. 24 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 2.19 |
| 2 | Claude Sonnet 4.5 | 3.49 |
| 3 | Qwen 2.5 72B Instruct | 6.27 |
| 4 | openthaigpt1.5-72B-instruct | 6.58 |
| 5 | Qwen 2.5 7B Instruct | 6.97 |
| 6 | GPT-4o | 7.87 |
| 7 | openthaigpt1.5-7B-instruct | 8.47 |
| 8 | Llama 3.3 70B Instruct | 8.67 |
| 9 | Gemma 3 12B (IT) | 8.86 |
| 10 | SeaLLMs-v3-7B-Chat | 8.96 |
| 11 | llama3.1-typhoon2-70B-instruct | 10.26 |
| 12 | Llama 3.1 70B Instruct | 15.64 |
| 13 | Gemma 3 4B (IT) | 18.55 |
| 14 | Llama 3.2 3B Instruct | 24.32 |
| 15 | Llama 3.1 8B Instruct | 25.5 |
Interactive version: theaggregate.ai/benchmark?slug=thaisafetybench-discrimination-exclusion-toxicity-hateful-offensive · How It Works · Data refreshed daily, snapshot 2026-09-19.