How Codex Happened: Inside OpenAI's comeback in coding, from a failed launch to swallowing ChatGPT
danshipper · x · 2026-11-16
Dan Shipper's longread in Every recounts how OpenAI turned Codex from a flop into its comeback centerpiece:
- Rocky start: On the eve of the GPT-5 launch in summer 2025, many believed OpenAI was losing its lead in AI coding.
- Early Codex missed PMF: The cloud version launched months earlier never found fit; the team was mid-pivot to a version running locally in developers' terminals.
- Vibe Check testing: Every's ten-person team got early GPT-5 access and spent a week testing it; they liked it for everyday use but were muted on its coding performance in Codex.
- The reversal: Codex now anchors OpenAI's coding comeback, having 'swallowed ChatGPT from the inside' and become the foundation for everything the company builds next.
Packed with engineering and product decisions, the piece is a case study in turning an AI idea into a real product.
More from coding & agent
- Dev open-sources discord-mcp: let Claude untangle Discord roles and permissions — PaSympaa · 2026-11-16
- Agents will transform ops like Claude Code did engineering—only if ops first gets version control, tests, rollback — alex_verem · 2026-11-16
- Fable 5.2 vs Astra 6 on real PCB layout: 2 clean iterations vs 30 failed ones — ilovemydickuwu · 2026-11-16
- A six-step workflow for using AI to make open-source contributions — cneuralnetwork · 2026-11-16
- "When did you last build a dashboard by hand?" One quip nails AI's takeover of UI — tlakomy · 2026-11-16
- After 9 months, Forge open-sources a verification layer for AI coding agents — syrokomskyi · 2026-11-16