Alibaba has just launched Qwen3, a new series of open source AI models that experts say rival Google and OpenAI’s top systems.
This is a bold move by China to increasingly dominate open source AI.
Qwen3 is not only another large language model. This is a hybrid inference system that can be fused with fast response With a slower, deliberate “thinking” mode, “no thought” allows developers to control performance and speed.
“We have seamlessly integrated thinking and non-thinking patterns,” Alibaba said.
Models range from 0.6B to 235B parameters, many models are now available for free download Hug the face,,,,, githuband Alibaba Cloud. The largest version, QWEN-3-235B-A22B, beat OpenaiO3 Minni and Google’s Gemini 2.5 Pro Regarding key coding and reasoning benchmarks, although not publicly available.
Inside AI
- Training after 36t token
- Supports 119 languages
- Use a “hybrid” (MOE) architecture to increase efficiency
- Built for everything from mobile AI to cloud deployment
Photography Solen Feyissa exist Unplash
Domination Contest
QWEN3 shows that China’s AI labs are rapidly closing the gap with American giants Open source. Although U.S. chip restrictions are designed to slow China’s progress, this is the latest upgrade in U.S.-China’s AI competition.
“Despite the pressure on U.S. export controls, this is a major breakthrough,” analyst Ray Wang said.
rise
Alibaba said the Qwen series has been downloaded more than 300 million times, with 100k+ derivative models on Hug Face. With the coming of DeepSeek’s R2, Baidu is also spinning towards open source, and China’s AI momentum is also difficult to ignore.
Qwen3 proves that cutting-edge AI is no longer from Silicon Valley, and the open source border may now be led by China.
