ChatGPT Work vs Codex: A Practical Task-Routing Guide
ChatGPT now has several work surfaces that can look interchangeable until you give them a real project. The routing rule I use is based on the output I need, not on which surface sounds more powerful.
Chat is best for questions, search, brainstorming, and fast judgment. Work is for longer research and multi-step deliverables that need to connect sources, files, and tools. Codex is for repository state: code changes, terminal commands, diffs, tests, debugging, and builds.
I tested that split on an operating Astro site. Chat made the initial editorial decision. Work connected OpenAI documentation, content inventory, SEO/GEO requirements, and project constraints. Codex then entered the repository, made the changes, reviewed the diff, and ran the build.
That workflow also exposed a useful failure mode: the first article version passed the technical checks but still failed editorial review. A green build proved the repository state was valid; it did not prove the content met the product standard.
The current product boundaries matter too. As of August 14, 2026, Cloud Work chats can continue across web, mobile, and desktop, while local desktop chats stay on that computer. Codex is still not a selectable web/mobile work surface, although supported desktop Codex conversations can be reached through the mobile Remote tab. Eligible desktop accounts can also use Voice with Work or Codex.
Those features make the surfaces feel closer, but the engineering boundary is still clear:
- use Chat when the bottleneck is deciding;
- use Work when the bottleneck is synthesizing and delivering;
- use Codex when the bottleneck is changing and validating a repository;
- keep human approval around delete/deploy/secret/production-write actions regardless of the surface.
Full case study and updated comparison:
Canonical: https://www.xbstack.com/en/ai/tools-lab/chatgpt-work-chat-codex-difference/
