Chinese tech firm Alibaba on Monday launched Qwen 3, a household of AI fashions the corporate claims matches and in some instances outperforms the very best fashions obtainable from Google and OpenAI.
Most of the fashions are — or quickly will probably be — obtainable for obtain beneath an “open” license from AI dev platform Hugging Face and GitHub. They vary in measurement from 0.6 billion parameters to 235 billion parameters. Parameters roughly correspond to a mannequin’s problem-solving expertise, and fashions with extra parameters usually carry out higher than these with fewer parameters.
The rise of China-originated mannequin sequence like Qwen have elevated the strain on American labs corresponding to OpenAI to ship extra succesful AI applied sciences. They’ve additionally led policymakers to implement restrictions geared toward limiting the flexibility of Chinese AI corporations to acquire the chips needed to coach fashions.
According to Alibaba, Qwen 3 fashions are “hybrid” fashions within the sense that they will take time and “cause” by means of advanced issues or reply less complicated requests shortly. Reasoning permits the fashions to successfully fact-check themselves, much like fashions like OpenAI’s o3, however at the price of increased latency.
“We have seamlessly built-in considering and non-thinking modes, providing customers the pliability to regulate the considering price range,” wrote the Qwen workforce in a weblog put up.
The Qwen 3 fashions help 119 languages, Alibaba says, and had been skilled on a knowledge set of almost 36 trillion tokens. Tokens are the uncooked bits of knowledge that the mannequin processes; 1 million tokens is equal to about 750,000 phrases. Alibaba says Qwen 3 was skilled on a mix of textbooks, “question-answer pairs,” code snippets, and extra.
These enhancements, together with others, vastly boosted Qwen 3’s efficiency in comparison with its predecessor, Qwen 2, says Alibaba. On Codeforces, a platform for programming contests, the biggest Qwen 3 mannequin — Qwen-3-235B-A22B — beats out OpenAI’s o3-mini. Qwen-3-235B-A22B additionally bests o3-mini on the most recent model of AIME, a difficult math benchmark, and BFCL, a check for assessing a mannequin’s means to “cause” about issues.
But Qwen-3-235B-A22B isn’t publicly obtainable — not less than not but.
The largest public Qwen 3 mannequin, Qwen3-32B, continues to be aggressive with quite a few proprietary and open AI fashions, together with Chinese AI lab DeepSeek’s R1. Qwen3-32B surpasses OpenAI’s o1 mannequin and Google’s Gemini 2.5 Pro on a number of checks, together with an accuracy benchmark known as LiveBench.
Alibaba says Qwen 3 “excels” in tool-calling capabilities in addition to following directions and copying particular knowledge codecs. In addition to releasing fashions for obtain, Qwen 3 is on the market from cloud suppliers together with Fireworks AI and Hyperbolic.
Tuhin Srivastava, co-founder and CEO of AI cloud host Baseten, mentioned that Qwen 3 is one other level within the development line of open fashions conserving tempo with closed-source programs corresponding to OpenAI’s.
“The U.S. is doubling down on limiting gross sales of chips to China and purchases from China, however fashions like Qwen 3 which can be state-of-the-art and open […] will undoubtedly be used domestically,” he informed TechCrunch in a press release. “It displays the fact that companies are each constructing their very own instruments [as well as] shopping for off the shelf by way of closed-model corporations like Anthropic and OpenAI.”
