Alibaba's Qwen3.7-Max Becomes First Chinese AI to Enter Top 4 in Coding Rankings (BABA)

Alibaba Group Holding Ltd.'s (BABA) Qwen3.7-Max model ranked fourth globally on the Code Arena benchmark, becoming the first Chinese-developed artificial intelligence to break into the top tier of the competitive programming evaluation. The flagship model scored 1,541 points as of May 24, 2026, trailing only Anthropic's Claude Opus 4.7 and Claude Opus 4.6 Thinking, while surpassing offerings from OpenAI and Google. The benchmark, considered a harsh test of real-world coding ability, evaluates front-end development, multi-step reasoning and agentic coding workflows. Qwen3.7-Max accumulated more than 1,522 votes with a narrow confidence interval, securing the highest rank among non-Anthropic models. Developer tests showed the model completed tasks with costs approximately 3.3 times lower than Opus 4.7 and 4 times lower than ChatGPT-5.5. Alibaba said the model is positioned as an agentic foundation model for long-duration autonomous tasks. Internal tests demonstrated a 35-hour continuous coding run with over 1,158 tool calls without context drift. The model is now available via Alibaba Cloud's Model Studio.