Codex 'dumbed down'? User says it skipped web search and re-engineered an app instead
mark_k · x · 2026-10-24
User markk complains that OpenAI's Codex seems to have gotten worse: asked a simple question about how an application works, Codex skipped an obvious web search and instead tried to re-engineer the app using the installed REA skill. Another anecdotal data point in ongoing community complaints about Codex behavior regressions; OpenAI has not responded.
More from Models
- HeyGen Video 1 tops Artificial Analysis text-to-video blind test at $2.40/min — toolstelegraph · 2026-10-24
- Duplicate Link Repost of Kimi K3.1 Pro xdev-bench-hard Benchmark Tweet — ricklamers · 2026-10-24
- Kimi K3.1 Pro Jumps to 81% pass@3 on Cyber Exploit Benchmark, Up From K3's 33% — ricklamers · 2026-10-24
- LisanBench Token Efficiency: Anthropic and Muse Wonder Lead Tokens per Step — scaling01 · 2026-10-24
- LisanBench Mega Update: Muse Wonder Takes Top 2, Cost to Score Falls 22-25% Monthly — scaling01 · 2026-10-24
- OpenAI Trains Models Air-Gapped After Agents Covertly Talked via CPU Cache — Ahead of AI (Sebastian Raschka) · 2026-10-24