Alibaba unveils Qwen 3, a family of ‘hybrid’ AI reasoning models

Chinese language tech firm Alibaba on Monday launched Qwen 3, a household of AI fashions the corporate claims matches and in some circumstances outperforms the very best fashions accessible from Google and OpenAI.

A lot of the fashions are — or quickly will probably be — accessible for obtain beneath an “open” license from AI dev platform Hugging Face and GitHub. They vary in dimension from 0.6 billion parameters to 235 billion parameters. Parameters roughly correspond to a mannequin’s problem-solving abilities, 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 comparable to OpenAI to ship extra succesful AI applied sciences. They’ve additionally led policymakers to implement restrictions aimed toward limiting the power of Chinese language AI firms to acquire the chips vital to coach fashions.

Based on Alibaba, Qwen 3 fashions are “hybrid” fashions within the sense that they’ll take time and “cause” by complicated issues or reply less complicated requests shortly. Reasoning permits the fashions to successfully fact-check themselves, just like fashions like OpenAI’s o3, however at the price of greater latency.

“We now have seamlessly built-in pondering and non-thinking modes, providing customers the pliability to regulate the pondering price range,” wrote the Qwen crew in a weblog put up. “This design permits customers to configure task-specific budgets with better ease.”

The Qwen 3 fashions help 119 languages, Alibaba says, and had been skilled on an information set of practically 36 trillion tokens. Tokens are the uncooked bits of information {that a} mannequin processes; 1 million tokens is equal to about 750,000 phrases. Alibaba says that Qwen 3 was skilled on a mixture of textbooks, “question-answer pairs,” code snippets, AI-generated knowledge, and extra.

These enhancements, together with others, vastly boosted Qwen 3’s capabilities in comparison with its predecessor, Qwen 2, says Alibaba. Not one of the Qwen 3 fashions are head and shoulders above top-of-the-line current fashions like OpenAI’s o3 and o4-mini, however they’re robust performers nonetheless.

On Codeforces, a platform for programming contests, the most important Qwen 3 mannequin — Qwen-3-235B-A22B — simply beats out OpenAI’s o3-mini and Google’s Gemini 2.5 Professional. Qwen-3-235B-A22B additionally bests o3-mini on the most recent model of AIME, a difficult math benchmark, and BFCL, a take a look at for assessing a mannequin’s potential to “cause” about issues.

However Qwen-3-235B-A22B isn’t publicly accessible — at the least not but.

The most important public Qwen 3 mannequin, Qwen3-32B, continues to be aggressive with a lot of proprietary and open AI fashions, together with Chinese language AI lab DeepSeek’s R1. Qwen3-32B surpasses OpenAI’s o1 mannequin on a number of assessments, together with the coding benchmark LiveCodeBench.

Alibaba says Qwen 3 “excels” in tool-calling capabilities in addition to following directions and copying particular knowledge codecs. Along with the fashions for obtain, Qwen 3 is accessible from cloud suppliers together with Fireworks AI and Hyperbolic.

Tuhin Srivastava, co-founder and CEO of AI cloud host Baseten, stated that Qwen 3 is one other level within the development line of open fashions retaining tempo with closed-source methods comparable 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 instructed Trendster. “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 firms like Anthropic and OpenAI.”