PANews reported on November 25th that, according to the Shanghai Securities News, the Artificial Intelligence Agency of Singapore (AISG) has abandoned the Meta Llama model in its latest Southeast Asian language proficiency model project, opting instead for Alibaba's Qwen architecture. To date, the Qwen series has been downloaded over 600 million times globally. It is reported that "Qwen-SEA-LION-v4," released by AISG on November 25th, ranks first in the Southeast Asian language proficiency rankings. Previously, open-source models, represented by Meta's Llama series, performed poorly when handling regional languages such as Indonesian, Thai, and Malay, severely limiting the development efficiency and performance of localized AI applications.PANews reported on November 25th that, according to the Shanghai Securities News, the Artificial Intelligence Agency of Singapore (AISG) has abandoned the Meta Llama model in its latest Southeast Asian language proficiency model project, opting instead for Alibaba's Qwen architecture. To date, the Qwen series has been downloaded over 600 million times globally. It is reported that "Qwen-SEA-LION-v4," released by AISG on November 25th, ranks first in the Southeast Asian language proficiency rankings. Previously, open-source models, represented by Meta's Llama series, performed poorly when handling regional languages such as Indonesian, Thai, and Malay, severely limiting the development efficiency and performance of localized AI applications.