How Codex Happened: Inside OpenAI's coding comeback from a failed launch
every · x · 2026-10-16
Dan Shipper of Every publishes a long-form oral history of OpenAI's comeback in AI coding. In summer 2025, his team got early GPT-5 access for a pre-launch review but found its coding underwhelming — at the time Codex was a scrappy effort whose cloud version had failed to find product-market fit and was mid-pivot to a terminal-based local version. Today Codex has become the centerpiece of OpenAI's strategy, effectively absorbing ChatGPT from within and serving as the foundation for everything the company builds next. The piece includes Mike Deal's early desk sketches for the article.
More from coding & agent
- Moving agent work from GPUs to CPUs for durable, predictable actions — cramforce · 2026-10-16
- Jev tutorial: 106x cheaper, 180ms yes/no gates replace PASS/FAIL string matching — max_paperclips · 2026-10-16
- Stop giving agents raw API keys: prepaid balances and per-credential caps — Cold_Hall_5384 · 2026-10-16
- Six-month prediction: hiring for generic agent harnesses built from scratch will cool — yi111 · 2026-10-16
- Omarchy ISO wouldn't boot on NVIDIA, so he had Claude patch it — indragie · 2026-10-16
- Dev uses Grok to clean up 3D printer firmware, catches it sneaking in nonsense code — yacineMTB · 2026-10-16