Alibaba Qwen 3.7: Faster AI Models, Global Impact

Alibaba has released new Qwen 3.7 previews to developers, intensifying the competition in AI models. Developers have spotted Qwen3.7-Max-Preview and Qwen3.7-Plus-Preview in Qwen Chat. This move comes as Qwen prepares for a "heavyweight new friend" announcement at Alibaba Cloud's May 20 summit. What's interesting is how quickly the Qwen team is moving. Qwen's models are known among developers for their throughput, memory use, and coding performance. They are being compared against a wide range of frontier AI systems, not just other Chinese models. The bigger story is that Qwen is becoming a benchmark challenger. The previous Qwen3.6-Plus focused on real-world agent performance, with stronger coding and multimodal reasoning. Developers now want models that can handle complex tasks and integrate with existing systems. Efficiency is also key. Recent tests showed significant throughput gains for Qwen3.6 on consumer hardware, with one setup jumping from 38 to 65 tokens per second. This means models can run comfortably and affordably. The bottom line is that the Qwen name is gaining influence beyond Alibaba's cloud. It offers companies a credible option for private AI deployments, especially for multilingual or locally controlled applications.