Search
A new wave of domestic AI large model releases is imminent, and Alibaba's Qwen Qwen4 series is showing impressive performance. In the GameFeel Bench evaluation, which focuses on game development creativity, Qwen4-Flash scored 49.9, surpassing Opus 5.5 xHigh's 47.2 and becoming the first domestic large model to lead Opus 5.5 in this evaluation. The local small model Qwen4-27B scored 44.6, close to Sonnet 5.5's 45. In terms of cost-effectiveness, Qwen4-Flash costs only $0.4 to complete tasks—far below Opus 5.5's $2.6—while Qwen4-27B costs only $0.16. The series includes at least four models; in addition to the above two, there are also Qwen4 Plus and Qwen4 Max, positioned at the upper limit of performance. The Qwen CTO previously promised to release the series before November 8, at which point real-world comparisons with mainstream overseas large models can be conducted.
-
C114 Communication Network -
Communication Home
