By Xiong Weisheng

BEIJING, Mar 28 (China Economic Net) — Beijing Academy of Artificial Intelligence (BAAI) and European open-source group Eclipse Foundation signed a cooperation agreement on Friday to develop an open-source large model evaluation project tailored to European requirements.

The agreement, signed at the Zhongguancun Forum in Beijing, will centre on PanEval, an open-source evaluation project that builds on BAAI's FlagEval framework and is designed for European use cases, according to information released at the event.

"Rather than replicating existing systems directly, PanEval adapts this foundation to address Europe's specific needs around governance, regulatory compliance, and linguistic diversity. This includes alignment with the EU AI Act, as well as a strong emphasis on data autonomy and transparency," the Foundation's head of AI Michael Berns said in a recent blog post.

Michael Plagge, VP Ecosystem Development at the Eclipse Foundation, said more cooperation with Chinese partners could follow, including in artificial intelligence and automotive-related projects.

Michael Plagge, VP Ecosystem Development at the Eclipse Foundation, in interview with China Economic Net [CEN/ Xiong Weisheng]

"There are a lot of things currently under discussion. There will be more coming," he said in an interview with China Economic Net (CEN) on the sidelines of the forum.

Asked about Chinese open-source and open-weight AI models in Europe, Plagge said language could in some cases be an obstacle. "Sometimes it feels to me it's a language barrier," he said, adding that Alibaba's Qwen was already among the most used models in Europe.

He also said some Chinese models appeared highly open by international standards, with developers in some cases releasing not only model weights but also training data.

The deal highlights a growing focus in cross-border AI cooperation on testing, governance and regulatory fit, as companies seek to adapt models and infrastructure for overseas markets.