OpenAI is retiring dated GPT-4o and GPT-4.1 versions from the API: what to change before 14 January 2027

If your product, chatbot, or automation still calls GPT-4o or GPT-4.1 through OpenAI’s API, you have a hard date. On 14 October 2026, OpenAI told developers that certain GPT-4 series models are deprecated and will be removed from the API on 14 January 2027. The recommended replacement for all of them is GPT-6 Luna (gpt-6-luna). Here’s what is shutting down, how this differs from the earlier ChatGPT retirement, and a practical checklist so small businesses don’t wake up to broken chats.

What is shutting down on 14 January 2027

OpenAI’s official deprecations page lists these snapshots, all with the same shutdown date and the same recommended replacement:

OpenAI also says fine-tuned versions of these models “follow the same shutdown schedule as their underlying base models.” OpenAI’s deprecations page says that at shutdown, a model “will no longer be accessible.”

What about the plain names? OpenAI’s model pages list plain names such as gpt-4o, gpt-4o-mini, gpt-4.1 and gpt-4.1-mini as aliases that point to a dated version (for example, gpt-4o currently points to gpt-4o-2024-08-06). The January 2027 table lists only the dated versions, and OpenAI doesn’t say there what will happen to the plain names. So check whether the plain name you call will move or stop, and plan to switch before 14 January 2027.

This is separate from an earlier batch. OpenAI’s April 2026 legacy list already set 23 October 2026 as the shutdown for another GPT-4o snapshot (gpt-4o-2024-05-13) and for gpt-4.1-nano. Don’t mix those dates with the January 2027 table above.

This is the API, not ChatGPT

In January 2026, OpenAI announced it would retire GPT-4o, GPT-4.1, GPT-4.1 mini and o4-mini from ChatGPT on 13 February 2026, and said at the time: “In the API, there are no changes at this time.” The 14 January 2027 notice is a separate, later API notice covering the dated GPT-4o and GPT-4.1 family versions listed above.

If you only use ChatGPT in a browser, this API notice may not change your day. If any of your software, a vendor integration, or a Zapier-style automation still names these model IDs, it will.

What OpenAI wants you to move to

For every snapshot in the January 2027 table, OpenAI recommends gpt-6-luna.

OpenAI describes GPT-6 Luna as its “most efficient model for focused, high-volume tasks.” In the OpenAI API it is available as gpt-6-luna. On OpenAI’s pricing pages, standard short-context rates for GPT-6 Luna are $0.10 per 1 million input tokens and $0.50 per 1 million output tokens (cached input is $0.01 per 1 million tokens, and OpenAI says prompts with more than 272K input tokens are priced at 2x input and 1.5x output for the full request). Treat those figures as OpenAI’s published list prices, and check the live pricing page before you budget.

OpenAI also says you can monitor shutdowns programmatically: the List models endpoint includes an optional shutdown_date field on each model when a date has been announced.

A simple migration checklist for small businesses

You don’t need a big engineering programme. You do need to find every place that still asks for the old models:

1. Search your code and configs for the old names. Look for the dated IDs above, plus plain names like gpt-4o, gpt-4o-mini, gpt-4.1 and gpt-4.1-mini (these currently point to the dated versions), and any fine-tune IDs built on those bases.
2. Ask your vendors. If a booking bot, helpdesk add-on, or “AI WhatsApp” tool runs on OpenAI under the hood, ask which model ID they pin, and whether they have a Luna migration date before 14 January 2027.
3. Switch the model ID to gpt-6-luna in a staging copy first. Keep the same prompts and tools, then compare answers on your real customer questions.
4. Re-check the answers you care about. Prices, hours, booking rules, refunds, and anything you allow an AI to say to a customer. A new model can still be wrong in new ways.
5. Decide what happens when the AI is unsure. Whatever model you use, the safe pattern for customer chat is the same: answer from information you approved, say when it doesn’t know, and hand over to a person.
6. Don’t wait until mid-January. OpenAI’s notice from 14 October 2026 to 14 January 2027 is about three months. That is enough time to test, but not if you start in the last week.

What this means for AI customer chat on WhatsApp

Model names change on a schedule you don’t control. Your customer still expects a fast, accurate reply on WhatsApp, and an honest handover when the bot is stuck.

At ThinkChat, we think the right AI CS hire answers from information you approved, stays honest when it doesn’t know, and can hand over to a person when it should. ThinkChat is not part of OpenAI’s deprecation programme, and we don’t claim any special migration path from GPT-4o. The checklist above is useful whichever provider you use.

Soft next step

If you want to hire a ThinkChat AI CS employee, Talk to Lyia on WhatsApp: +60 11-5515 9318.

Sources

Checked 16 October 2026.

• OpenAI, “Deprecations” (entry “2026-10-14: GPT-4 series model deprecations”; also the “2026-04-22: Legacy GPT model snapshots” entry) (same page at https://platform.openai.com/docs/deprecations). Accessed 16 Oct 2026.
• OpenAI, model pages for GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini (snapshots and aliases), https://developers.openai.com/api/docs/models/gpt-4o-mini, https://developers.openai.com/api/docs/models/gpt-4.1, https://developers.openai.com/api/docs/models/gpt-4.1-mini. Accessed 16 Oct 2026.
• OpenAI, “GPT-6 Luna” model page. Accessed 16 Oct 2026.
• OpenAI, “Pricing”. Accessed 16 Oct 2026.
• OpenAI, “Introducing GPT-6 Sol and Luna” (page carries an update note dated 29 September 2026)
• OpenAI, “Retiring GPT-4o, GPT-4.1, GPT-4.1 mini, and OpenAI o4-mini in ChatGPT” (29 January 2026)