Computer Use is Human-like Now, But Still Slow
anpaure · x · 2026-07-14
This reply debates whether AI computer use capabilities are overhyped. The author feels that "only Blender seems legit," likely because the model just outputs instructions in batches, almost like writing a manual.
They add that other computer use scenarios are still "painfully slow." However, compared to a few months ago, models can now perform mouse and keyboard operations at an APM speed approaching that of a StarCraft player, sometimes even crashing the browser.
Related event: LLM Computer Use Skills Show Rapid Leap(3 posts)→
More from coding & agent
- Claude adds screen-recorded skills that can replay your workflow — CodeByPoonam · 2026-07-22
- Devin adds e2b sandboxes for remote agent execution — badphilosopher · 2026-07-22
- Hermes Agent Refactoring Proposal: Decoupling via Event Bus and Monorepo Slicing — Promptmethus · 2026-07-22
- ty now reads Pydantic config keywords and field metadata — charliermarsh · 2026-07-22
- Pensar Launches AI Security Agent to Autonomously Discover and Patch 0-Days — andriy_mulyar · 2026-07-22
- ty adds first-class Pydantic support, including strict and lax field handling — charliermarsh · 2026-07-22