xbench, developed by HongShan Capital Group, introduces a novel AI benchmark focusing on real-world task execution alongside standard testing. Unlike many static benchmarks, xbench features an ever-evolving suite of tests to enhance evaluation rigor and adaptability. The initiative seeks to promote open-source development within the AI community, bringing forth two recently open-sourced benchmarks, xbench-Science QA and xbench-DeepSearch. Future updates will continuously reflect advancements in AI and large models, enhancing the relevance of these benchmarks in assessing AI capabilities.
"xbench evaluates models not only on the ability to pass arbitrary tests but also on the ability to execute real-world tasks, which is more unusual."
"HSG stated that its original intention in creating xbench was to attract more AI talents and projects in an open and transparent way."
Collection
[
|
...
]