The Qwen3 series includes two so-called mixture-of-experts (MoE) models that are trying to match hybrid reasoning systems - which mimic the way humans think through problems - recently introduced by Anthropic and Alphabet's Google.
Alibaba Group Holding took the wraps off a new version of its flagship Qwen AI model, sustaining the breakneck pace of development that's characterised China's artificial intelligence sphere in the wake of DeepSeek.
Alibaba on Tuesday unveiled its Qwen3 series of models, which it said rivals DeepSeek's performance on several fronts, including math and coding. Qwen3 also significantly cuts deployment costs compared to other major models, the company said.
