close
close

Alibaba makes Sora video models Open Source, as the tech giant AI functions shows

Alibaba Cloud, The Cloud Computing and Ai Arm from Alibaba, said on Wednesday that the four models of the WAN2.1 series – made the latest version of his video foundation model Tongyi Wanxiang – for academics, researchers and commercial institutions worldwide. Alibaba has the South China Morning Post.

The four models are trained on 1.3 or 14 billion parameters, said Alibaba Cloud and added that they are named in their own AI model community, as a model scope, as well as the collaborative AI platform for the hugging of the face.

The open source approach for offering software is gaining in dynamics, partly thanks to the deep hango-based Deepseek, the Chinese Ki start-up, which is its popular major language models open sources. The trend has in turn initiated a wave of AI applications between Chinese companies and consumers.

Open sourcing offers public access to the source code of a software program with which third-party developers can change or release their design, repair broken links or scale its functions. The use of the approach has gained a rapid introduction according to the popularity of deepseek models, which will keep up with the performance of OpenAis models to a fraction of the costs.

The WAN2.1 series published in January was the first model of video generation to support text effects in both Chinese and English. The model is now the top video generation model for VBohrch Bohrlochboard, a benchmark suite for video models on the hug face. It is the only open source model among the top 5 on the list.