We seriously need benchmark for research, wide web search and fact retrieval.

A Reddit user highlights the critical need for a benchmark to evaluate research, wide web search, and fact retrieval capabilities. They emphasize that such a benchmark should be tested in actual environments, allowing access to tools and unrestricted internet searches, similar to systems like Claude or ChatGPT Work. The user is actively seeking information on whether such a benchmark currently exists.

This post highlights the absence of a benchmark for research and fact retrieval that allows unrestricted tool and internet access, unlike existing benchmarks with limitations.