Chinese language expertise firm Alibaba launched on Thursday new open-source synthetic intelligence fashions and text-to-video AI expertise, intensifying its efforts to compete within the booming space of generative AI.
The open-source fashions, numbering greater than 100, come from Alibaba’s Qwen 2.5 household, its newest foundational massive language mannequin launched in Could.
Chinese language expertise corporations, like their U.S. counterparts, have been investing closely in generative AI, with companies racing to develop sturdy product portfolios and diversified choices.
Whereas rivals akin to Baidu and OpenAI have primarily adopted closed-source approaches, Alibaba has embraced a hybrid mannequin, investing in each proprietary and open-source improvement to broaden its AI product vary.
Its new fashions vary from 0.5 to 72 billion parameters – the variables that decide an AI mannequin’s functionality and efficiency – in measurement, providing proficiency in arithmetic, coding, and help for over 29 languages, Alibaba mentioned in a press release.
The fashions purpose to cater to a wide selection of AI functions throughout varied sectors together with automotive, gaming, and scientific analysis.
Alibaba on Thursday additionally unveiled a brand new text-to-video mannequin as a part of its Tongyi Wanxiang picture era household, becoming a member of a rising variety of Chinese language tech companies coming into this rising market.
The transfer places Alibaba in direct competitors with international gamers akin to OpenAI, which has additionally proven curiosity in text-to-video expertise.
In August, ByteDance, proprietor of TikTok, launched its text-to-video app Jimeng AI on Apple’s App Retailer for Chinese language customers.
© Thomson Reuters 2024
