openthaigpt1.5-72B-instruct: benchmark results

Provider: Other. Released 2024-09-30. Access: Open.

Unified ELO 1664 ± 27, rank #419 of 2656 rated models, from 35 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
Thai LLM MC - m3exam_tha_seacrowd_qa70.39Accuracy (%)100
Thai LLM MC - thaiexam_qa63.89Accuracy (%)97.8
Thai LLM NLU - belebele_tha_thai_seacrowd_qa88.78Accuracy (%)97.1
Thai LLM NLU - xcopa_tha_seacrowd_qa95.4Accuracy (%)97.1
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (BLEU)53.03BLEU97
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (chrF++)61.96chrF++97
Thai LLM - Coding7.95Rating (0-10)95.8
Thai LLM - Math8.05Rating (0-10)95.8
Thai LLM NLU - xnli.tha_seacrowd_pairs46.25Accuracy (%)94.2
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (BLEU)29.18BLEU94
Thai LLM NLG - flores200_eng_Latn_tha_Thai_seacrowd_t2t (SacreBLEU)40.3SacreBLEU94
Thai LLM NLG - flores200_tha_Thai_eng_Latn_seacrowd_t2t (SacreBLEU)37.41SacreBLEU94

Interactive version: theaggregate.ai/model?slug=openthaigpt1-5-72b-instruct · How It Works · Data refreshed daily, snapshot 2026-09-19.