OpenAI's messy execution vs Anthropic's principles: a Reddit take blames the executives' backgrounds
thntu · reddit · 2026-10-05
A Reddit user argues OpenAI's chronic execution problems — the GPT 4/5 naming mess, fragmented Codex/chat/work modes, tightened usage limits with PR-style resets, coding models a step behind Anthropic (6.1 Sol under Opus 5.5), and rumors of peeking at user chats — trace back to leadership: OpenAI's execs are entrepreneurs/engineers, while Anthropic's leaders (Amodei in biophysics and RLHF research, Kaplan in physics and scaling laws) have serious research backgrounds. The author is switching to Claude this month but watching OpenAI's promised improvement plan.
More from Companies & People
- Fund manager who hired Zain at 14 backs Ghost AI founder ahead of Core launch — jamdac · 2026-10-06
- Tesla did adopt Waymo-style geofenced FSD, but nationwide unsupervised rollout still hasn't happened — binarybits · 2026-10-06
- Tesla still hasn't delivered nationwide unsupervised FSD promised within two years — binarybits · 2026-10-06
- Manus is hiring Technical Account Managers in SF and Singapore — parker_lyman · 2026-10-06
- Overheard on a flight raving about Factory, a CEO turned it into a hire — maddiehfaulkner · 2026-10-06
- Agent skills are saturated—go deeper into MLOps, inference engineering, RL — kmeanskaran · 2026-10-06