Alibaba’s Enhanced Qwen3 Surpasses OpenAI and DeepSeek – A Look at Alibaba Group Holdings (NYSE:BABA)

Alibaba Group (BABA) has upgraded its Qwen3 family of large language models (LLMs), introducing the Qwen3-235B-A22B-Instruct-2507-FP8. This new model showcases enhanced AI capabilities in key areas such as math, coding, logical reasoning, and instruction adherence, outshining competitors like OpenAI and DeepSeek in benchmarks. Notable features include support for inputs of up to 256,000 tokens—an eightfold increase—allowing for longer text processing. Additionally, the 3-billion-parameter Qwen model will be integrated into HP’s “Xiaowei Hui” smart assistant, streamlining document drafting and meeting summarization. Performance metrics indicate that Qwen3-Coder surpassed domestic rivals, matching performance levels of U.S. models like GPT-4. Nvidia’s CEO recently lauded the Qwen series as one of the leading AI systems globally, reflecting China’s growing prominence in the AI sector. Alibaba’s stock has surged over 42% this year, significantly outperforming the NYSE Composite Index.

Source link