GPT-4o API deprecation: GPT-4o and 4.1 shut down on January 14, 2027
GPT-4o API deprecation: on January 14, 2027 OpenAI shuts down GPT-4o, GPT-4o mini and GPT-4.1. Check if it affects you and how to move to the new model.
This article in other languages:العربيةDeutschEspañolFrançaisItaliano日本語Polski简体中文
In this article
The GPT-4o API deprecation is now official: on January 14, 2027, OpenAI will shut down GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini in the API, along with their fine-tuned versions. After that date, requests to these models will stop working. If your chatbot, automation or integration uses any of them, you have three months to migrate, and OpenAI points to gpt-6-luna as the replacement.
What OpenAI announced
On October 14, 2026, OpenAI notified developers that several GPT-4 series models are being retired. The notice is on the official API deprecations page. The shutdown covers these versions:
| Model | API identifier | Recommended replacement |
|---|---|---|
| GPT-4o | gpt-4o-2024-08-06, gpt-4o-2024-11-20 |
gpt-6-luna |
| GPT-4o mini | gpt-4o-mini-2024-07-18 |
gpt-6-luna |
| GPT-4.1 | gpt-4.1-2025-04-14 |
gpt-6-luna |
| GPT-4.1 mini | gpt-4.1-mini-2025-04-14 |
gpt-6-luna |
OpenAI states plainly that models fine-tuned on these bases will be shut down on the same date as the base models.
It helps to remember what “deprecation” means in OpenAI’s vocabulary: a model becomes deprecated the moment it is announced, and on the shutdown date it stops being available. There is no grace period after that date.
An earlier deadline: October 23, 2026
That is not the only date you need in your calendar. As soon as October 23, 2026, OpenAI shuts down older models announced for retirement in April, including gpt-3.5-turbo, gpt-4, gpt-4-turbo, gpt-4.1-nano, gpt-4o-2024-05-13, o1, o3-mini and o4-mini. If you have an old integration somewhere that was never updated, you have a week, not three months.
Does the GPT-4o API deprecation affect your company
It does if you use the OpenAI API directly, or through a tool that calls these models under the hood. GPT-4o mini and GPT-4.1 mini are cheap models that were easy to plug into email classification, document data extraction or a simple chatbot, and then forget about. Integrations like that can run for years without anyone touching the code.
The quickest way to check:
- Search your code and configuration for
gpt-4oandgpt-4.1. Pay attention to environment variables and settings in no-code tools (Make, Zapier, n8n). - Open the usage dashboard on the OpenAI platform and see which models are generating traffic.
- Ask the vendors of off-the-shelf tools which model they run on and when they plan to migrate.
How to move to gpt-6-luna
According to the model page, gpt-6-luna is “our most efficient model for focused, high-volume tasks”. It has a 1,050,000-token context window. API pricing is $0.10 per million input tokens and $0.50 per million output tokens. For comparison, GPT-4o mini costs $0.15 and $0.60, and GPT-4.1 mini $0.40 and $1.60. On paper, then, the migration does not raise your bill.
Two things I would watch out for:
- Reasoning is on by default. The
reasoning.effortparameter ingpt-6-lunadefaults tomedium. For simple tasks, testnoneorlow. That should cut response time and the number of output tokens, but check it on your own data. - Long prompts cost more. Above 272K input tokens, the price for the whole request goes up (2× for input, 1.5× for output).
My advice: don’t just swap the model name in your config. Collect 50–100 real examples from production, run them through the old and the new model, and compare the results. A newer model doesn’t always answer in the same format, and JSON parsers and prompts written for GPT-4o can break on that.
Fine-tuning: the least time
If you have a model fine-tuned on GPT-4o mini or GPT-4.1, things are harder. According to the May 7, 2026 entry, active customers will lose the ability to create new fine-tuning jobs on January 6, 2027. That is eight days before the base models shut down. If you still need fine-tuning, schedule training on the new model with plenty of margin, or check whether a good prompt with examples gets you similar results without fine-tuning.
TL;DR
When will GPT-4o disappear from the OpenAI API?
On January 14, 2027. This applies to gpt-4o-2024-08-06 and gpt-4o-2024-11-20, as well as GPT-4o mini, GPT-4.1 and GPT-4.1 mini.
What should replace GPT-4o mini?
OpenAI recommends gpt-6-luna. According to OpenAI’s pricing, it is cheaper than GPT-4o mini on both input and output.
What happens to my fine-tuned model?
It will be shut down on the same day as its base model, January 14, 2027. New fine-tuning jobs can only be created until January 6, 2027.
Does this affect ChatGPT?
This announcement is about the API. GPT-4o already disappeared from ChatGPT on February 13, 2026.