BrowseComp: a benchmark for browsing agents

OpenAI releases BrowseComp benchmark for evaluating AI agents' performance on complex multi-step web browsing tasks.

Categories: Research

Read at source: https://openai.com/index/browsecomp