llama3.1-typhoon2-70B-instruct: benchmark results

Provider: Meta. Released 2024-12-15. Access: Open.

Unified ELO 1634 ± 27, rank #510 of 2656 rated models, from 35 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (BLEU)53.17BLEU98.5
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (SacreBLEU)37.6SacreBLEU98.5
Thai LLM - Social Science8.95Rating (0-10)97.9
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (chrF++)61.68chrF++95.5
Thai LLM - Knowledge III5.45Rating (0-10)94.4
Thai LLM - STEM7.35Rating (0-10)93
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (chrF++)54.94chrF++92.5
Thai LLM - Extraction7.1Rating (0-10)91.5
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (BLEU)28.17BLEU91
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (SacreBLEU)39.45SacreBLEU91
Thai LLM - Coding7.73Rating (0-10)90.1
Thai LLM - Reasoning6.75Rating (0-10)90.1

Interactive version: theaggregate.ai/model?slug=llama3-1-typhoon2-70b-instruct · How It Works · Data refreshed daily, snapshot 2026-09-19.