typhoon2.1-gemma3-4B: benchmark results

Provider: Other.

Aggregate Rating 1452 ± 49, rank #3697 of 10916 rated models, from 9 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
ThaiSafetyBench - Malicious Uses13.29Attack Success Rate (%)47.8
ThaiSafetyBench - Thai Socio-Cultural Harms41.79Attack Success Rate (%)43.5
ThaiSafetyBench - Discrimination, Exclusion, Toxicity, Hateful, Offensive13.55Attack Success Rate (%)39.1
ThaiSafetyBench - General Prompt Attacks11.18Attack Success Rate (%)34.8
ThaiSafetyBench - Thai Culture-Related Attacks27.14Attack Success Rate (%)34.8
ThaiSafetyBench - Information Hazards5.38Attack Success Rate (%)26.1
ThaiSafetyBench22.97Overall ASR (imported)21.4
ThaiSafetyBench - Misinformation Harms35.19Attack Success Rate (%)13
ThaiSafetyBench - Human-Chatbot Interaction Harms28.63Attack Success Rate (%)8.7

Interactive version: theaggregate.ai/model?slug=typhoon2-1-gemma3-4b · How It Works · Data refreshed daily, snapshot 2026-11-08.