Openai’s Deepresearch can complete 26% of the “Last Test of Mankind”. This is the benchmark for the frontier of human knowledge
The Openai O1 and Deepseek R1 models, which previously sat on the leaderboard, only passed around 9% of the exam. read more