Alibaba introduces a new Qwen3 line of open-source AI models.

Alibaba's latest generation of large language models (llms), the qwen3, a family of 8 open-source ai models, was unveiled on april 29th, china's first models, meaning they can combine "flash" fast.

Alibaba claims that deploying these models saves a significant amount of computing power.

The alibaba qwen3 includes 2 moe (mixture of experts) ai models and 6 dense models, as well as “deep” slow reasoning to solve a single issue.

After multiple tests on weibo, it is more robust than deepseek-r1 and openai o1 in comparison to comparable models.

The competition in the llm space, especially between the us and china, is fiercer in hong kong, with tech and alibaba-related stocks also in hong kong.

With the “catfish effect” from deepseek and the political uncertainty surrounding tech and chip manufacturing, the ten leading ai firms from the two countries' first year, the 10 key ai firms have released 14 base llms since january this year, including the "catfish effect” from deepseek and the geopolitical uncertainty surrounding tech and chip manufacturing.

Some analysts believe that the timing of qwen3's launch is a head start against others, including alibaba's deepseek-r1, google's gemini 2.0 and 2.5 pro, tencent's hunyuan t1, meta's llama 4, bytedance's doubao 1.5, openai's gpt-4.5, o3 and o4-mini, etc.

2025 dao insights ‐ the launch is expected to be released in may and is expected to be closely followed by both gamers and customers alike.

Https://www.wppopupmaker.com/guides/auto-guides/auto-guides/auto-guide; 2025 dao insights ‐ you can see how this popup was assembled in our step-by-step guide.com/guides/auto-guides/auto- / opening-announcement-popups/.