Google Research has published a blog post describing tests of large language models on questions related to superconductivity research. The post is categorized under Education Innovation, suggesting the focus is on how such models might support learning or research in a highly technical domain.
Because the source is a single blog post, there are no contrasting sources to compare. The supplied source does not include specific methodology, model names, or quantitative results, so the article can only confirm the existence and general direction of the testing effort.
Readers should treat the post as an early look at an ongoing evaluation rather than a definitive assessment of LLM performance in superconductivity research.