Comment by visarga

The thig is, if you look at all the "Deep Research" benchmark scores, they never claim to be perfect. The problem was plain to see.