Critic Concedes OpenAI's GPT Computer-Use Now Drives Safari Like a Human, Barely Errs
kimmonismus · x · 2026-10-05
kimmonismus, who has recently written many critical posts about OpenAI's decisions, gives credit where it's due: GPT models' computer-use capabilities have become phenomenal. The model drives Safari as if a human were controlling the browser, makes hardly any mistakes, and carries out tasks independently well enough to genuinely surprise him.
More from coding & agent
- Codex tip: sidechat busts your prompt cache and eats usage limits fast — brandon_galang · 2026-10-05
- Dev claims mystery system hits 100% on agentic benchmarks, can't explain how — examachine · 2026-10-05
- AI coding tools could be breaking the junior engineer pipeline — Suspicious_Orchid770 · 2026-10-05
- Meta built its internal agent platform on MCP, packaging expertise as reusable skills — Suspicious_Orchid770 · 2026-10-05
- Hugging Face Engineer Shows Browser-Run WebAI Doing Search, Speech and Background Removal Without the Cloud — nicodotdev · 2026-10-05
- mitsuhiko and badlogic on why they built Pi Durable instead of using Temporal — mitsuhiko · 2026-10-05