Artificial Analysis finds Qwen4-Max lags Claude Opus 5.5

thenewstack.io7 min read

An independent test comparing Alibaba's Qwen4-Max with Anthropic's Claude Opus 5.5 found that while Qwen4-Max costs about a third of Opus, it takes over three times as long and makes more accuracy mistakes on complex agent tasks. The author concludes that Qwen4-Max is well-suited for high-volume data investigation work where cost matters and extra time is acceptable, but recommends keeping Opus for coding and tasks requiring high reliability. Open article →