Qwen3.5 is 60pc cheaper to make use of and eight instances higher at processing massive workloads, the corporate stated.
Alibaba has unveiled its newest AI mannequin known as Qwen3.5, as newer launches from Chinese language firms catch as much as their US counterparts within the race for AI dominance.
The primary mannequin within the open-weight Qwen3.5 collection demonstrates “excellent outcomes throughout a spread of benchmarks”, the corporate stated. It ranks larger than OpenAI’s GPT-5.2, Anthropic’s Claude Opus 4.5 and Google’s Gemini 3 Professional in a number of of the assessments.
The mannequin is constructed on a hybrid structure that enables solely 17bn parameters to activate per ahead move, whereas comprising of a complete of 397bn parameters. This, Alibaba stated, optimises pace with out sacrificing its functionality.
In response to the corporate, Qwen3.5 is 60pc cheaper to make use of and eight instances higher at processing massive workloads than its fast predecessor. The brand new mannequin comes with “visible agentic capabilities”, Alibaba stated – the power to take actions throughout cellphone and pc apps.
“Constructed for the agentic AI period, Qwen3.5 is designed to assist builders and enterprises transfer sooner and do extra with the identical compute, setting a brand new benchmark for functionality per unit of inference value,” the corporate stated in a press release as reported by Reuters.
Alibaba’s newest launch follows ByteDance, which launched an upgraded model of its Doubao chatbot app over the weekend. The agentic chatbot service has near 200m customers.
The TikTok-parent additionally launched the most recent model of its AI video generator known as Seedance 2.0, which garnered reward for its means whereas additionally receiving criticism for potential copyright theft.
Different Chinese language AI leaders launched their very own new fashions lately, together with Zhipu, which unveiled GLM-5, skilled totally utilizing Chinese language chips, MiniMax, which launched M2.5, and the Alibaba-backed Moonshot AI, which got here out with Kimi K2.5.
These new launches come forward of DeepSeek’s new V4 mannequin, anticipated to return out later this month. In response to studies, the brand new DeepSeek mannequin may outperform rivals ChatGPT and Claude, notably on duties that contain lengthy coding prompts.
Don’t miss out on the data it is advisable succeed. Join the Every day Temporary, Silicon Republic’s digest of need-to-know sci-tech information.
Elevate your perspective with NextTech Information, the place innovation meets perception.
Uncover the most recent breakthroughs, get unique updates, and join with a worldwide community of future-focused thinkers.
Unlock tomorrow’s developments right this moment: learn extra, subscribe to our e-newsletter, and grow to be a part of the NextTech group at NextTech-news.com
