Alibaba Releases Qwen3.7-Max AI Model for Autonomous Tasks

Alibaba’s Qwen team has introduced Qwen3.7-Max, a proprietary AI model designed to perform complex, autonomous tasks over multiple days [1]. The model, which debuted on the Arena AI leaderboard in May 2026, marks a strategic shift for the company as it prioritizes monetizing its most advanced technology through paid APIs while offering less performant versions as open source [1, 3].

Key takeaways

• Qwen3.7-Max demonstrated the ability to execute autonomous engineering tasks for 35 continuous hours, completing over 1,000 tool calls [1].
• The model is integrated into vehicles from several major automakers, including BYD, Volkswagen, and Geely, to provide voice-activated digital services [2].
• Alibaba now ranks as the sixth-ranked AI lab globally for text-based models [3].
• Access to Qwen3.7-Max is limited to Chinese-based endpoints, which may impact its adoption by Western enterprises concerned with data sovereignty [1].

Advancing Autonomous Agent Capabilities

Qwen3.7-Max is built as a "versatile agent foundation" intended to overcome the tendency of language models to degrade during long-horizon reasoning tasks [1]. In a controlled test, the model autonomously optimized an attention kernel on an unfamiliar server architecture, achieving a 10.0x speedup over 35 hours [1]. Beyond technical coding, the model has demonstrated capabilities in simulating startup lifecycles, navigating decision-making rounds, and performing autonomous office tasks like document reformatting [1].

The model’s architecture supports "cross-harness generalization," allowing it to function as a drop-in intelligence layer for frameworks like Claude Code [1]. While Alibaba has positioned the Max version as a premium, proprietary product, it remains competitively priced compared to Western frontier models, costing $2.50 per million input tokens and $7.50 per million output tokens [1].

Automotive Integration and Market Expansion

Beyond enterprise software, Alibaba is aggressively integrating Qwen into the automotive sector. During the 2026 Beijing Auto Show, the company announced that its AI would power in-car digital services for a wide range of manufacturers, including SAIC Volkswagen, Li Auto, and Great Wall Motor [2]. These systems utilize a combination of on-device processing and cloud computing to allow drivers to perform tasks such as booking hotels, ordering food, and tracking packages via voice commands [2]. This rollout is part of a broader effort by Chinese automakers to differentiate their vehicles in a slowing electric vehicle market through advanced software [2].

Why it matters

The release of Qwen3.7-Max highlights the industry’s transition into the "agent era," where AI models are expected to plan and execute multi-step workflows without constant human intervention [1]. By aligning its business model with American labs like and , Alibaba is attempting to recoup the high costs of training large-scale models through paid subscriptions [1]. However, the model’s restriction to Chinese endpoints and the departure of key Qwen team leaders earlier this year remain factors that could influence its long-term global adoption and competitive standing [1].