Alibaba's Qwen4 series outperforms opus 5.5 in game development creativity test, offers significant cost savings

Alibaba's Qwen4 model series has recently demonstrated a significant technological leap, surpassing the leading overseas model Opus 5.5 in a specialized creativity evaluation for game development, known as the GameFeel Bench. Specifically, the lightweight Qwen4-Flash achieved a score of 49.9, exceeding Opus5.5xHigh's score of 47.2, marking the first instance where a Chinese large model has surpassed a rival of this stature in this benchmark. Furthermore, the smaller, locally deployable Qwen4-27B scored 44.6, performing nearly on par with Sonnet5.5's score of 45. Beyond performance, the Qwen4 series presents a substantial economic advantage. Leaked data indicates that Qwen4-Flash requires only about $0.4 to execute an equivalent task compared to Opus5.5's $2.6, while Qwen4-27B's operational cost is even lower at $0.16, positioning it as highly competitive within its tier.