ANALYSIS

Alibaba Releases Third Closed-Source AI Model in Three Days as It Pivots to Profit

M MegaOne AI Apr 2, 2026 3 min read
Engine Score 5/10 — Notable
Editorial illustration for: Alibaba Releases Third Closed-Source AI Model in Three Days as It Pivots to Profit
  • Alibaba Group released its third proprietary AI model in three consecutive days, signaling an aggressive shift toward monetizing its flagship Qwen AI services.
  • The move marks a strategic pivot from Alibaba’s earlier open-source approach, where it freely released Qwen model weights to build developer adoption.
  • Alibaba’s cloud division, which houses its AI efforts, reported its first operating profit in 2025 after years of losses.
  • The rapid-fire release cadence puts Alibaba in direct commercial competition with Baidu, ByteDance, and other Chinese AI providers charging for API access.

What Happened

Alibaba Group Holding Ltd. released its third proprietary AI model on April 2, 2026, continuing a streak of daily closed-source launches that began on March 31. The releases come through Alibaba’s cloud computing division, which develops and hosts the company’s Qwen family of AI models. According to Bloomberg, the company is reinforcing its intent to profit from its AI services rather than give them away.

Alibaba had previously built its reputation in AI by open-sourcing several versions of its Qwen models, which gained significant traction among developers and researchers globally. The shift to closed-source, paid models represents a fundamental change in strategy.

Why It Matters

Alibaba’s pivot mirrors a broader trend in the Chinese AI industry, where companies that initially competed on open-source model releases are now racing to convert developer interest into revenue. Baidu’s Ernie Bot and ByteDance’s Doubao have already established paid API tiers, and Alibaba’s move intensifies competition in the Chinese enterprise AI market.

The timing is significant. China’s AI market has been characterized by intense price wars throughout 2025, with providers repeatedly undercutting each other on API pricing. Alibaba’s decision to release multiple closed-source models in rapid succession suggests the company believes it has enough differentiation to charge premium prices.

Technical Details

Alibaba has not disclosed full technical specifications for the three new closed-source models. The Qwen model family has historically spanned a range of parameter counts, from the compact Qwen2.5-0.5B to the full Qwen2.5-72B. The company’s previous open-source releases scored competitively on standard benchmarks including MMLU, HumanEval, and GSM8K.

The closed-source models are expected to be available through Alibaba Cloud’s Model Studio platform, which provides API access with usage-based pricing. Alibaba Cloud reported processing over 1.6 trillion API calls for its Qwen models during the fourth quarter of 2025.

Who’s Affected

Developers and companies that relied on Alibaba’s open-source Qwen models may face pressure to evaluate whether the new closed-source offerings justify paid access. Enterprise customers in China and Southeast Asia who use Alibaba Cloud for AI workloads are the primary target market. Competitors including Baidu, ByteDance, and Tencent will need to respond to Alibaba’s expanded commercial AI lineup.

What’s Next

Alibaba is expected to detail pricing and capability breakdowns for the new models at its upcoming Alibaba Cloud Summit. The company’s cloud division will report quarterly earnings in May, providing the first concrete data on whether the closed-source strategy is translating into AI revenue growth. Open-source versions of Qwen remain available, but future releases may increasingly be reserved for the paid tier.

Share

Enjoyed this story?

Get articles like this delivered daily. The Engine Room — free AI intelligence newsletter.

Join 500+ AI professionals · No spam · Unsubscribe anytime

M
MegaOne AI Editorial Team

MegaOne AI monitors 200+ sources daily to identify and score the most important AI developments. Our editorial team reviews 200+ sources with rigorous oversight to deliver accurate, scored coverage of the AI industry. Every story is fact-checked, linked to primary sources, and rated using our six-factor Engine Score methodology.

About Us Editorial Policy