- New flagship model places fifth worldwide in Artificial Analysis rankings
- Alibaba positions latest Qwen release for AI agent and enterprise workloads
Alibaba’s newly released Qwen3.7-Max model ranked fifth globally and first among Chinese AI models in the latest benchmark published by independent evaluation platform Artificial Analysis, underscoring intensifying competition between Chinese developers and leading US frontier-model firms.
According to the rankings released on May 21, Qwen3.7-Max scored 56.6 points, improving 4.8 points from Alibaba’s previous flagship release and outperforming domestic rivals including Moonshot AI’s Kimi-K2.6, DeepSeek’s DeepSeek-v4-Pro-Max and Zhipu AI’s GLM5.1.
The model trailed leading systems from major US AI developers, including OpenAI’s GPT-5.5, Anthropic’s Claude-Opus4.7 and Google’s Gemini 3.1 Pro Preview.
Alibaba Cloud is expected to make Qwen3.7-Max available through API access on its Bailian AI platform.
Alibaba said the model was specifically designed for AI agents and enterprise-grade autonomous workflows, with improvements in coding, reasoning and tool-use capabilities.
The company said Qwen3.7-Max can integrate with agent frameworks including Claude Code, OpenClaw, Hermes Agent and Qwen Code, allowing it to autonomously handle extended tasks involving programming and repeated tool invocation.
Alibaba said the model is capable of independently executing long-duration workflows lasting up to 35 hours and involving more than 1,000 tool calls.
Artificial Analysis is widely followed within the AI industry as an independent benchmarking platform evaluating large models across multiple performance categories.
Alibaba’s Qwen family has repeatedly ranked near the top of the platform’s leaderboard. The company’s earlier Qwen3.6-Max-Preview release, launched roughly one month ago, had previously held the highest ranking among Chinese-developed models.
