AISG has officially switched to Alibaba’s Qwen architecture for its language proficiency project, leaving behind Meta’s Llama model due to its inefficiencies with regional languages.
The new Qwen-SEA-LION-v4 model, introduced on November 25th, has achieved the highest position in language proficiency rankings across Southeast Asia.
With over 600 million global downloads, the Qwen model series demonstrates significant acceptance and usage in the tech community.
Meta’s Llama model was inadequate for regional languages such as Indonesian and Thai, hindering effective localized AI applications.
Leave a Reply