Alibaba has unveiled its latest AI innovation, the Qwen3.5, a sophisticated large language model (LLM) that merges traditional capabilities with advanced dynamic reasoning. This release comes as the company navigates stiff competition within China’s booming artificial intelligence landscape, where multiple AI models were recently launched.
The Qwen3.5 model is available in two formats: an open-weight version, which permits users to download, run, fine-tune, and deploy the model on their own systems, and a hosted version that operates on Alibaba’s servers. Both variants became available on the eve of the Chinese New Year, coinciding with another AI model launch tailored for robotics just a week earlier. According to Alibaba, Qwen3.5 boasts enhancements in performance and cost efficiency, alongside what the company describes as “native multimodal capabilities.” This feature allows the model to process text, images, and video concurrently.
In alignment with a prominent trend in AI development, Qwen3.5 supports new coding and agentic features, making it compatible with open-source AI agents like those developed by OpenClaw, which have gained significant traction recently. These agents can autonomously execute tasks and manage complex actions with minimal human oversight. The heightened interest in these capabilities follows the release of new agent tools by American AI firm Anthropic, reshaping expectations in the market regarding the potential of AI agents to replace traditional software as a service companies.
The Qwen3.5 model is distinguished by its substantial architecture, comprising 397 billion parameters-elements that determine how the system learns and reasons. While this is fewer parameters than its predecessor, Alibaba asserts that the new model has shown significant improvements based on self-reported benchmarks. The company claims that Qwen3.5’s performance holds up against leading models from OpenAI, Anthropic, and Google DeepMind, although these comparisons were made by Alibaba itself.
In addition to its capabilities, Qwen3.5 supports a total of 201 languages and dialects, a notable increase from the previous generation’s offering of 82 languages. Lin Junyang, the technical lead for Alibaba Cloud’s Qwen team, indicated that more open-weight models are expected to be rolled out during the Chinese New Year period.
Following Anthropic’s introduction of its Claude AI agent tools, American tech giants have been ramping up their efforts in developing agent capabilities. OpenAI CEO Sam Altman announced that a creator from OpenClaw will be joining the firm, indicating a strategic shift in the competitive landscape. Additionally, Google DeepMind’s Demis Hassabis has commented that Chinese AI models are rapidly closing the gap on their Western counterparts, suggesting an ever-evolving dynamic in the global AI race.
As Alibaba continues to navigate this competitive environment, Qwen3.5 emerges as a pivotal player, highlighting the company’s commitment to advancing AI technology while responding to market demand.
Original Source: https://www.cnbc.com/2026/02/17/china-alibaba-qwen-ai-agent-latest-model.html
Category :
Tags:
Publish Date: 2026-02-17 08:50:00
