OpenAI Bought a Whole Company So Codex Could Code 25 Hours After You Close Your Laptop
Last Updated on June 18, 2026 by Editorial Team
Author(s): Chew Loong Nian – AI ENGINEER
Originally published on Towards AI.
Why an agent dies when you close the laptop
OpenAI documented a single Codex run that went for about 25 hours uninterrupted, burned roughly 13 million tokens, and produced around 30,000 lines of code without a human touching the keyboard. That number sat in a research note for months as a curiosity. Last week it stopped being a curiosity.
The article argues that OpenAI’s acquisition of Ona (formerly Gitpod) is less about improving coding models and more about enabling long-running agents to survive and keep working after a developer closes their laptop. It explains why local agent processes die (sleep/network loss), contrasts that with cloud sandboxes that keep jobs running, and frames the key metric as “task horizon” (how long an agent can complete work unattended). It summarizes evaluation trends showing rapidly increasing long-horizon performance, then outlines how to start with Codex CLI and use long-horizon “goal” mode safely. The piece then covers the automation layer (scheduled/event-driven agent runs), and finally the tradeoffs nobody screenshots: cost/limits, drift and failure modes during long autonomy, and the need for supervision, configuration (like AGENTS.md), monitoring, and guardrails. It concludes with practical guidance on when to use local vs cloud modes and reframes success as better task specification and review rather than higher short-task benchmark scores.
Read the full blog for free on Medium.
Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.
Published via Towards AI
Towards AI Academy
We Build Enterprise-Grade AI. We'll Teach You to Master It Too.
15 engineers. 100,000+ students. Towards AI Academy teaches what actually survives production.
Start free — no commitment:
→ 6-Day Agentic AI Engineering Email Guide — one practical lesson per day
→ Agents Architecture Cheatsheet — 3 years of architecture decisions in 6 pages
Our courses:
→ AI Engineering Certification — 90+ lessons from project selection to deployed product. The most comprehensive practical LLM course out there.
→ Agent Engineering Course — Hands on with production agent architectures, memory, routing, and eval frameworks — built from real enterprise engagements.
→ AI for Work — Understand, evaluate, and apply AI for complex work tasks.
Note: Article content contains the views of the contributing authors and not Towards AI.
Related posts
Recent Posts
