@_can1357: did you notice how claude occasionally runs commands that take hours, but GPT never does? all because codex prefixes to…
Summary
The tweet points out that Claude occasionally runs commands that take hours while GPT does not, due to Codex prefixing tool results with wall time for RL environments to convey time understanding.
View Cached Full Text
Cached at: 09/15/26, 11:44 AM
did you notice how claude occasionally runs commands that take hours, but GPT never does? all because codex prefixes tool results with wall time so the RL env gives the model some understanding of time as a construct; claude doesn’t. funny isn’t it?
Similar Articles
I ran Claude code and Codex in parallel for 15 days. Here's what I found.
The author reviews Tutti, a shared multi-agent workspace, after using it for 15 days with Claude Code and Codex, highlighting workflow improvements like the @ system and agent borrowing, along with some drawbacks.
@RoundtableSpace: THIS GUY FIGURED OUT HOW TO RUN CLAUDE CODE, CODEX, AND CURSOR INSIDE GROK BOT IN JUST 4 MINUTES https://x.com/Av1dlive…
A user discovered a method to run AI tools like Claude Code, Codex, and Cursor inside Grok Bot in just four minutes, claiming it represents the closest thing to AGI they have experienced.
@VraserX: Tibo seems to be hinting that Codex may have overtaken Claude Code. If that number is usage or retention, that’s genuin…
Tibo hints that OpenAI's Codex may have overtaken Anthropic's Claude Code in usage or retention metrics, suggesting heightened competition among AI coding agents.
Claude Code vs Codex: 37 min but working UI, vs 27 min but broken twice
A comparison of Claude Code and Codex in a refactoring task reveals that Claude took longer but produced more thorough code with better UI design, while Codex was faster but required additional prompts to work properly.
@DavidKPiano: Been using Codex much more than Claude Code lately It's wild that Claude became popular *because* it had the best-in-cl…
David K. Piano notes he's been using OpenAI's Codex more than Anthropic's Claude Code recently, suggesting that Claude's coding lead may be slipping.