Work I used to give juniors now takes Opus 5.5 minutes. I don't think we've clocked how far past GDPval we are

A developer notes that tasks previously assigned to junior staff now take Opus 5.5 minutes to complete. This observation follows reports that GPT-5.2 achieved 71% success against industry professionals in December, with GPT-5.5 reaching 85% and Opus 4.7 80% by April. OpenAI subsequently ceased publishing human-graded performance metrics. The developer is curious how others are integrating AI into their daily office work, specifically if they are delegating more tasks to AI for review rather than direct execution.