Alibaba to Release Upgraded Qwen 3 AI Model in Late April

Alibaba Group Holding reportedly plans to launch an upgraded model of its flagship AI mannequin, Qwen 3, later this month.

The timing of the discharge might change, Bloomberg reported Tuesday (April 1), citing an unnamed supply.

Alibaba didn’t instantly reply to PYMNTS’ request for remark.

The corporate’s newest launch will come at a time when the competitors within the discipline is heating up, in keeping with the Bloomberg report.

Chinese language corporations have launched a number of low-cost synthetic intelligence (AI) companies since DeepSeek gained consideration with the discharge of its mannequin that it mentioned was inexpensive to develop than these of its American rivals, the report mentioned.

OpenAI, Google and Anthropic have additionally launched new AI fashions in current weeks, per the report.

Alibaba itself has already launched some AI merchandise this 12 months, together with a brand new mannequin in its Qwen 2.5 collection final week and a brand new model of its AI assistant Quark final month, in keeping with the report.

The corporate mentioned in February that it’ll spend extra on AI within the subsequent three years than it has within the final decade.

Alibaba administration mentioned throughout a Feb. 20 earnings name that the corporate’s AI investments are based mostly on the first aim of reaching synthetic common intelligence (AGI).

“We intention to proceed to develop fashions that stretch the boundaries of intelligence,” Eddie Wu, the corporate’s CEO, mentioned throughout the earnings name. “Why is that the first intention? Properly, it’s as a result of all the seen AI software situations in the present day that we see round content material creation, search and so forth and so forth have arisen exactly on account of the continued extension of these boundaries, and we need to maintain pushing out these boundaries to create increasingly more alternatives.

When Alibaba Cloud introduced the launch of the brand new Qwen2.5-Omni-7B on Thursday (March 27), it mentioned this multimodal AI mannequin can course of inputs within the type of textual content, pictures, audio and video; can generate real-time responses within the type of textual content and pure speech; and will be deployed on cell phones and laptops.

As a result of the mannequin is each compact and multimodal, it could possibly energy “agile, cost-effective AI brokers,” the corporate mentioned.