TL;DR: Alibaba Qwen's new flagship model tops the Code Arena: WebDev leaderboard, beating Claude Opus 5 and Kimi K3 while staying on the cost-efficiency Pareto frontier at blended $5/MToken.
Summary: Qwen3.8-Max-0902 from Alibaba Qwen scored 1691 points in the Code Arena: WebDev benchmark, taking #1 overall — 3 points above Claude Opus 5 (Max) and 17 above Kimi K3 (Max). The model also ranks #1 in Data & Analytics and Consumer Product roles, with top-3 placements across all six WebDev categories. Priced at a blended $5/MToken, it claims the highest-scoring Pareto frontier position among the tested models.
Why it matters: AI builders building web-facing products now have a new high-performance, low-cost frontier coding model to evaluate. Watch released Agent Arena scores and benchmark Qwen3.8-Max-0902 against your own web dev and data-dashboard workflows.
Source: x_com