Researcher Seeks Human Baselines for Screenspot-Pro Agent Evaluations
——0——
Sentiment
Pos100%
100%
50%
Neg0%
Users are pleasantly surprised that Opus 4.7 oneshotted the ScreenSpot-Pro webapp right out of the box because of its strong unexpected performance on the agent evaluation task.
1 comment with sentiment.
57111.2K
Cluster engagement
18 snapshots
Researcher Seeks Human Baselines for Screenspot-Pro Agent Evaluations · Digg
