Codex 'gpt-5.5 is not supported' error: what to use now

Quick answer: Codex says "The 'gpt-5.5' model is not supported when using Codex with a ChatGPT account" because OpenAI retired GPT-5.5 from Codex for ChatGPT sign-ins on 14 October 2026 at noon PDT (19:00 UTC), and something in your setup still asks for it. On Plus, Pro, Business, Enterprise and Edu, switch to GPT-6.1 Sol (gpt-6.1-sol). On Free and Go, choose GPT-6 Luna (gpt-6-luna) in the desktop app. Then replace gpt-5.5 everywhere it's saved: config files, workspace defaults, custom agents, scheduled tasks and scripts. The OpenAI API and Codex sessions that use an API key aren't affected.

Key facts

• Retired: GPT-5.5 in Codex with ChatGPT sign-in, on 14 October 2026 at 12:00 PDT (19:00 UTC).
• Not affected: the OpenAI API and Codex sessions authenticated with an API key.
• Plus, Pro, Business, Enterprise and Edu: use GPT-6.1 Sol. On Enterprise and Edu it stays off until an admin enables it.
• Free and Go: use GPT-6 Luna in the desktop app, where it has rolled out.
• Codex CLI 0.161.0 (7 October) already made GPT-6.1 Sol the default model in its bundled catalog.
• Earlier Codex retirements: gpt-5.4 and gpt-5.4-mini on 31 August (ChatGPT sign-in), GPT-5.3-Codex-Spark on 14 September.

Why you see this error

Codex sent a request for gpt-5.5 while you're signed in with ChatGPT, and since 14 October that combination isn't accepted. OpenAI's models page lists where to replace it: workspace defaults, saved model settings, managed configurations, custom agents, scheduled tasks and scripts that select a model. A leftover gpt-5.5 in any of them can cause the error, even if you never picked it yourself. Per the config docs, Codex reads settings in this order, highest first: command-line flags and --config overrides, project .codex/config.toml files, profile files, your user config at ~/.codex/config.toml, cloud-managed workspace defaults and the system config.

How to fix it

• CLI: run /model and pick GPT-6.1 Sol, or start Codex with codex --model gpt-6.1-sol.
• Config file: in ~/.codex/config.toml, change model = "gpt-5.5" to model = "gpt-6.1-sol". Check any .codex/config.toml inside your repos and any profile files (~/.codex/profile-name.config.toml) too.
• Desktop app and IDE extension: choose a new model in the model picker and check saved model settings.
• Admins: update workspace defaults and managed configurations. On Enterprise and Edu, enable GPT-6.1 Sol first.
• Custom agents, scheduled tasks and scripts: replace gpt-5.5 wherever a model is set, including scripts that pass --model or -m.
• Free and Go: pick GPT-6 Luna in the desktop app.
• Old CLI: versions from 0.161.0 on default to GPT-6.1 Sol. Our install and update guide has the commands.

Which model to use instead

OpenAI recommends GPT-6.1 Sol for complex coding and agentic work and GPT-6 Luna for focused, repeatable tasks. Its pricing page estimates these local messages per five hours on Plus and Standard Business:

• GPT-6.1 Sol: 15 to 160.
• GPT-6 Sol: 15 to 150.
• GPT-6 Astra: 5 to 45.
• GPT-6 Luna: 350 to 3,000.

Pro plans currently have no five-hour limit, weekly limits may also apply, and OpenAI calls the numbers estimates, not fixed limits. GPT-6 Sol is still in Codex, but GPT-6.1 Sol is newer and is the model OpenAI names as the replacement. We compare the two in GPT-6 Sol vs GPT-6.1 Sol.

Old desktop threads can stay stuck

Two open bug reports on OpenAI's Codex GitHub, #54030 and #54226, describe the same error in the Windows desktop app. A long thread that started on GPT-5.5 keeps failing after it's switched to a current model, because the automatic compaction of its old history still calls GPT-5.5. The reporters say switching models, restarting the app and changing settings didn't help, and as of 18 October neither issue had a fix from OpenAI. These are user reports, not something OpenAI has confirmed. Until it's fixed, start new work in a new thread.

If you use an API key

The retirement only covers ChatGPT sign-in. OpenAI says the OpenAI API and Codex sessions authenticated with an API key aren't affected, so if you run Codex with an API key, this change doesn't apply to you.