Most contemporary AI models are not engineered for sustained, multi-step autonomous execution. Tasks such as executing hundreds of iterative code modifications or chaining tool calls over several hours without human intervention require a distinct model architecture and training focus. At the 2026 Alibaba Cloud Summit on May 20, the Qwen team officially announced Qwen3.7-Max.
Two preview versions of the series had previously appeared on the Arena AI leaderboard without a formal press release. According to LM Arena, Qwen3.7-Max-Preview ranked 13th globally in text capabilities (placing Alibaba as the #6 lab), while Qwen3.7-Plus-Preview ranked 16th in vision (placing Alibaba as the #5 lab). Qwen3.7-Plus-Preview is described as a high-performance balanced version focusing on reasoning and multimodal inputs, whereas Qwen3.7-Max is the text-only reasoning flagship now formally available via API.
The team describes Qwen3.7-Max as its most comprehensive agent model to date. This proprietary, closed-weight model is capable of handling complex coding, debugging, office workflow automation, and long-horizon tasks spanning hundreds or even thousands of steps. A key feature is the "Extended-Thinking Mode," where the model generates an internal chain of thought to plan, verify, and correct course before committing to a final answer.
Reasoning models like this produce significantly more output tokens than standard models. In evaluations by Artificial Analysis, Qwen3.7-Max generated approximately 97 million tokens, compared to a benchmark average of 24 million. While this overhead adds latency for simple tasks, it is the primary strength for multi-step planning and code refactoring where complex agentic chains are required.
The model features a massive 1M-token context window, a significant upgrade from the 256K window in Qwen3.6 Max Preview. This allows for processing entire mid-sized code repositories or large document stacks in a single request. While pricing for the new model has yet to be announced, the previous Qwen3.6 Max Preview was priced at $1.30/$7.80 per million input/output tokens on Alibaba Cloud.
