"Outsider of the Red Packet War"
```
Author | Chai Xuchen
Editor | Zhou Zhiyu
As the Spring Festival approaches, "cash giveaways" to promote new business have become a standard routine for major internet companies. This year, during the Spring Festival season, the main character in the red envelope war has become AI.
Recently, leading AI companies have successively announced events regarding Spring Festival cash red envelope campaigns. However, on the other side of the battlefield, Alibaba's Tongyi Qianwen has remained "low-key." They did not release guides for grabbing red envelopes, nor marketing gimmicks like "Let the dragon roar," but instead, on January 26, dropped a nuclear bomb on the global tech sector: the Qianwen flagship reasoning model Qwen3-Max-Thinking.
In today's traffic-saturated environment, using the huge traffic window of the Spring Festival and leveraging subsidy strategies to gain users’ downloads and daily active numbers, overtaking peers in terms of consumer user scale—this strategy is rife with the familiar smoke of the mobile internet era, and players are clearly trying to transplant this classic playbook onto the arena of large models.
Yet, some industry insiders point out that the underlying logic of the AI era has already diverged from the traffic rules of the mobile internet era. Large models are neither food delivery nor ride-hailing platforms—they are not simple matchmaking platforms, but rather productivity tools that must provide extremely high intellectual density.
An executive at a modeling company told Wallstreetcn that subsidies could previously retain users because the demand for services was homogeneous; but in the new era, if the model is not smart enough, if it cannot solve complex problems, then the masses of users lured in by red envelopes may eventually be lost.
The Alibaba team has clearly realized that the moat of the AI era is no longer user scale, but "the height of intelligence." Therefore, at this point on January 26, they chose not to jump into the marketing melee, but to release Qwen3-Max-Thinking.
This model is not only an iteration of past technologies, but also a dimensional reduction strike against the old logic of internet traffic. With its trillion-level parameters and profound reasoning capability, it demonstrates that the only true value in the AI arena is IQ, not DAU.
The release of Qwen3-Max-Thinking marks that China’s large models have formally moved beyond the "basic competition" for traffic, advancing into the deep water zone of competing for depth of thought.
According to information published by the officials, Qwen3-Max-Thinking's total number of parameters has exceeded the trillion mark, and it has undergone training with even larger-scale reinforcement learning. This method is no longer simply "feeding data," but uses proprietary algorithms to allow the model to engage in self-play and self-evolution amidst massive, complex tasks.
This pursuit of technological extremes has enabled Qwen3-Max-Thinking to surpass international top models like GPT-5.2, Claude Opus 4.5, and Gemini 3 Pro in multiple key performance benchmark tests, proving that building the intelligent high ground of computing power and algorithms can only be achieved by a steadfast, direct sprint.
Even more critically, the "Thinking" ability emphasized by Qwen3-Max-Thinking subverts the traditional logic of internet products.
In the past, internet products pursued speed—quick connections, fast matching, rapid response. In the AI era, Qwen3-Max-Thinking introduces a "slow thinking" mechanism similar to human System 2. When facing complex problems, it doesn’t respond instantly like a search engine via keyword matches; instead, it engages in multi-step reasoning, logical breakdown, and self-reflection.
This "thoughtful deliberation" trait may not match simple retrieval in response speed, but its value density grows exponentially. This is the core competitiveness that users truly pay for in the AI era.
Moreover, Qwen3-Max-Thinking’s breakthrough in Agent capability further illustrates the chasm between the new and old eras.
In the internet era, tools were fragmented and users had to jump between different apps, whereas Qwen3-Max-Thinking has greatly enhanced its native Agent ability to autonomously invoke tools, allowing the model to "think while using tools" like a professional.
It is no longer just a passive Q&A box, but a super assistant capable of proactively planning, real-time adjustment, and calling various APIs to complete complex tasks. This more user-friendly, smarter, and smoother interactive experience thoroughly breaks the boundaries of traditional internet applications.
Simultaneously, overcoming the persistent "hallucination" issue of large models also demonstrates Alibaba’s pragmatic attitude of solving real problems. By reducing hallucinations, Qwen3-Max-Thinking lays the foundation for tackling real-world complex tasks, which can gain the long-term trust of both business and consumers far more than any marketing gimmick.
The Spring Festival of 2026 may be the last curtain call for the old internet era’s way of thinking. Users will ultimately realize that those who truly change their lives and improve their efficiency are not cash rewards but the quietly operating, trillion-parameter brain in the background that can think deeply like a human. This is the true victory of the AI era.
Risk Reminder and DisclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice, nor does it consider the individual user’s investment objectives, financial situation, or needs. Users should consider whether any opinions, perspectives, or conclusions in this article are suitable to their own circumstances. Any investment made accordingly is at your own risk.
```