Codex still set to gpt-5.5? It retires Oct 14 under ChatGPT sign-in, so switch your config now

#What changed

On October 14, 2026, GPT-5.5 retires from ChatGPT, ChatGPT Work, and Codex on all plans, including consumer, Business, Enterprise, and Edu. OpenAI's models page says this retirement does not apply to the OpenAI API, where GPT-5.5 stays available.

#Who is affected

Anyone who uses Codex signed in with a ChatGPT account (desktop app, CLI, or IDE extension) and still selects gpt-5.5 somewhere: the model entry in config.toml, workspace defaults, saved model settings, managed configurations, custom agents, scheduled tasks, or scripts such as codex exec --model gpt-5.5. Workspace admins who manage model defaults for a team are affected too.

#What to do now

• On Plus, Pro, Business, Enterprise, and Edu, switch to GPT-6 Sol (gpt-6-sol), the replacement the docs name. The same page recommends GPT-6.1 Sol (gpt-6.1-sol) for complex coding where your account has it.
• On Free and Go, switch to GPT-6 Luna (gpt-6-luna) in the desktop app when it is available to you.
• Search your repos and dotfiles for "gpt-5.5" and update every hit, including CI jobs and scheduled tasks that would otherwise fail quietly.
• The docs give no time of day for the cutoff, so treat October 13 as your working deadline.
• Admins: check the workspace model availability guidance the docs link to, since newer models may need to be enabled for Enterprise and Edu workspaces.

#What is not confirmed

• OpenAI has not published a time of day for the retirement.
• The current models page states only that the OpenAI API is unaffected. OpenAI's September 15 announcement also said Codex sessions authenticated with an API key keep GPT-5.5, but the docs page does not repeat that for this retirement, so DotsFeed has not independently confirmed it.
• Earlier guidance suggested GPT-5.6 Sol or GPT-6 Astra as replacements; the current docs name gpt-6-sol and gpt-6-luna instead. Follow the current docs.

Spotted on X via @RowWhisper's reminder; details checked against OpenAI's models documentation.