NEWS

Alibaba Launches Qwen 3.5 — Claims to Outperform GPT-5.2

Alibaba Cloud unveils Qwen 3.5, a 397-billion-parameter Mixture-of-Experts AI model with native multimodal capabilities, claiming 60% lower costs and...

Dylan H.

News Desk

February 17, 2026
2 min read
Alibaba Launches Qwen 3.5 — Claims to Outperform GPT-5.2

China's AI Race Intensifies

Alibaba Cloud has unveiled Qwen 3.5 on February 16, positioning it as the company's most ambitious AI model to date. Designed for the agentic AI era, the model features a Mixture-of-Experts (MoE) architecture that activates only a fraction of its parameters per query, dramatically reducing inference costs.


Model Specifications

SpecificationDetails
Total Parameters397 billion
Active Parameters17 billion per forward pass
ArchitectureMixture-of-Experts (MoE)
ModalitiesText, images, video (up to 2 hours)
Cost Reduction60% lower than predecessors
Performance8x better than previous generation

Performance Claims

Alibaba asserts Qwen 3.5 outperforms the leading models from Western competitors:

  • OpenAI GPT-5.2
  • Anthropic Claude Opus 4.5
  • Google Gemini 3

These claims have not yet been independently verified by third-party benchmarks. Community evaluations on platforms like Chatbot Arena and Artificial Analysis are expected in the coming weeks.


Agentic Capabilities

Qwen 3.5 is specifically designed for autonomous agent workflows:

  • Tool use and function calling with improved reliability
  • Multi-step planning across complex task chains
  • Native code execution and debugging
  • Long-context reasoning across extended documents and conversations

Market Context

The launch comes amid intensifying competition in China's AI market, where Alibaba trails ByteDance's Doubao chatbot in domestic usage. Released on the eve of Chinese New Year, the timing is designed to maximize visibility.

The model also escalates the broader US-China AI rivalry. Chinese AI labs — including Alibaba, DeepSeek, and Baidu — have been closing the gap with Western counterparts, often at significantly lower development costs.


What to Watch

  • Independent benchmark results on Chatbot Arena and LMSYS
  • DeepSeek V4 — expected to launch around the same timeframe with 1 trillion parameters
  • Enterprise adoption across Alibaba Cloud's customer base
  • Open-weight availability — whether Alibaba releases model weights publicly

Qwen 3.5 represents another significant milestone in China's AI development. Whether its performance claims hold up under independent evaluation remains to be seen.