Alibaba Unveils Qwen3.5 with Visual Agentic Capabilities and Lower Costs
Alibaba has unveiled its latest AI model, Qwen3.5, as Chinese companies intensify competition with their US counterparts in the race for AI dominance. Notably, the launch signals continued momentum from China’s leading technology players.
The first open weight model in the Qwen3.5 series demonstrates “outstanding results across a full range” of benchmarks, the company said. In several tests, it ranks higher than OpenAI’s GPT-5.2, Anthropic’s Claude Opus 4.5 and Google’s Gemini 3 Pro.
Moreover, the model runs on a hybrid architecture that activates only 17bn parameters per forward pass, although it comprises 397bn parameters in total. As a result, it optimises speed while maintaining strong overall capability.
Lower Costs and Greater Efficiency
According to the company, Qwen3.5 is 60pc cheaper to use and eight times better at processing large workloads than its immediate predecessor. In addition, the model introduces “visual agentic capabilities” – the ability to take actions across phone and computer apps.
“Built for the agentic AI era, Qwen3.5 is designed to help developers and enterprises move faster and do more with the same compute, setting a new benchmark for capability per unit of inference cost,” the company said in a statement, as reported by Reuters.
Therefore, the new release targets developers and enterprises that seek higher performance without significantly increasing compute costs.
Intensifying Competition in China’s AI Sector
Meanwhile, the launch follows ByteDance releasing an upgraded version of its Doubao chatbot app over the weekend. The agentic chatbot service has close to 200m users.
In addition, the TikTok parent recently introduced Seedance 2.0, the latest version of its AI video generator. While many praised its performance, critics raised concerns about potential copyright theft.
At the same time, other Chinese AI leaders rolled out new models. Zhipu unveiled GLM-5, which it trained entirely using Chinese chips. MiniMax released M2.5, while the Alibaba-backed Moonshot AI introduced Kimi K2.5.
These developments come ahead of DeepSeek’s anticipated V4 model, which is expected later this month. According to reports, the upcoming model could outperform rivals such as ChatGPT and Claude, particularly on tasks involving long coding prompts.
