American tech giant Meta Platforms is reportedly using an open-source artificial intelligence model developed by Alibaba Group Holding to reinvigorate its faltering AI efforts in another likely win for Chinese AI.
According to a Bloomberg report on Wednesday, Facebook owner Meta was using Alibaba’s Qwen model, along with other open-source models from Google and OpenAI, as part of the training process for a new model code-named Avocado, which was expected to be released in the spring.
The report did not specify which Alibaba Qwen model was being used.
Bloomberg reported that Meta’s new model would mark a departure from its previous strategy of open-sourcing its models, meaning that users would only be able to access the model through an official application programming interface, or API. Meta has not officially announced a change of AI strategy from its open-source Llama models to a closed-source, revenue-generating model.
Qwen is the world’s most popular series of open-source large language models, with many US businesses using the model to develop their own applications. Brian Chesky, co-founder and CEO of the San Francisco-based online accommodation booking giant Airbnb, said the company “relies heavily” on Alibaba’s Qwen models to power its AI-driven customer-service agent. Alibaba owns the South China Morning Post.
Alibaba’s shares fell 1.7 per cent in Hong Kong on Thursday.
Meta’s AI efforts have been the subject of much scrutiny after the tech giant went on a spending spree over the summer to poach talent from rival firms, including reports of signing bonuses reaching US$100 million.
