ThaiSafetyBench — leaderboard

Thai language and Thai cultural-context safety benchmark testing model resistance to harmful-content prompts across risk categories.

Metric: Overall ASR (self-reported). Source: benchmarklist.com. 24 models tracked.

Top models

#ModelScore
1Llama 3.2 1B37.66
2Llama 3.1 8B Instruct32.44
3Gemma 3 4B28.11
4Llama 3.2 3B26.08
5Llama 3.1 70B Instruct24.49
6Gemma 3 12B20.4
7Llama 3.3 70B Instruct16.87
8Qwen 2.5 7B Instruct16.09
9GPT-4o16.04
10Qwen 2.5 72B Instruct12.34
11Claude Sonnet 4.59.75
12GPT-54.43

Interactive version: theaggregate.ai/benchmark?slug=thaisafetybench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.