Alibaba launches Qwen4 Max AI with 60% lower task costs
Alibaba has launched Qwen4-Max, its most advanced artificial intelligence model, with 2.4 trillion parameters and a claimed 60% cut in average cost per task. The Hangzhou-based company says the closed flagship activates fewer parameters than its predecessor while adding cybersecurity and CBRN safety upgrades, positioning Qwen4 against U.S. rivals.
The release puts Alibaba deeper into the global race with closed-source U.S. labs such as OpenAI and Anthropic. U.S.-listed BABA shares still edged more than 1% lower in overnight trading after the news, according to market reports.
Key Takeaways
- Alibaba debuted Qwen4-Max as its most advanced AI model to date.
- The 2.4-trillion-parameter system activates 25% fewer parameters per token than Qwen3.8-Max and claims a 60% drop in average cost per task.
- Qwen4-Max is closed, with smaller open-weight Qwen4 models promised later for developers.
- Safety work includes cyber guardrails plus datasets and post-training aimed at chemical, biological, radiological, and nuclear risks.
- The model is live on QwenCloud at 7.5 U.S. cents per 1 million tokens.
What did Alibaba announce with Qwen4?
According to company claims relayed in market coverage, Qwen4-Max matches its predecessor's 2.4 trillion total parameters but activates only 71 billion parameters per token—about 25% less than Qwen3.8-Max. Alibaba says that design improves reasoning while lowering inference costs.
The flagship is positioned for complex work tasks on an internal benchmark the company cited. Access is through QwenCloud, with pricing listed at 7.5 U.S. cents per 1 million tokens. For more Future Tech & AI Wonders coverage, browse BlasterPost's category hub.
How does Qwen4 cut costs versus earlier models?
Alibaba says activating fewer parameters per token lets Qwen4-Max deliver stronger reasoning at lower run cost, cutting average cost per task by 60% on that internal work-task benchmark. That efficiency pitch matters as Chinese and U.S. AI developers compete to raise capability without making models prohibitively expensive to operate.
Coverage framed the launch as part of a broader push to match U.S. rivals on performance while keeping usage economics competitive for enterprises and developers.
Why are Qwen4 safety upgrades in the spotlight?
Reports note stringent cybersecurity guardrails as Alibaba balances safety with its broader open-source strategy. The company introduced dedicated datasets and post-training enhancements meant to mitigate chemical, biological, radiological, and nuclear risks.
The launch lands amid debate over whether open AI models preferred by many Chinese firms could raise those risks more than closed systems from U.S. rivals. Qwen4-Max itself is closed, though Alibaba reiterated that smaller-scale open-weight Qwen4 versions would be released for global developers soon. Read the Reuters report for additional wire context on the safety framing.
How did markets react to the Qwen4 launch?
Despite the AI headline, BABA was down more than 1% overnight, and the stock is down more than 39% in 2026. Retail chatter on Stocktwits was described as bullish with high message volume, with some users praising Alibaba's AI agents while others questioned the selloff.
Taken together, Qwen4 is Alibaba's clearest recent answer on cost, capability, and safety as it vies with U.S. AI leaders—while keeping the top model closed for now and promising open-weight siblings later.