OpenAI will switch off GPT-4o and GPT-4.1 in its API on 14th Jan'27, with half its usual notice

Plenty of production software still runs on GPT-4o and GPT-4.1, because they were the default when it was built and nobody has had a reason to touch it since. Three months is enough time to change a model name but not much time to learn what changes with it, since a different model reads the same prompt differently and the output your code or your team has learned to expect will drift. Fine-tuned models are the harder case, because whatever that training was doing has to be rebuilt or replaced. Start by searching your code for those model names, and ask your vendors too, because some of them will sit inside tools you bought rather than built. Run a sample of real inputs through Luna and compare the answers before December, while there is still time to fix the prompts. OpenAI has also cut its own six-month promise in half, so treat any notice period as a hope and keep the model name in one place you can change.