OpenAI gives selected GPT-4 API models a January 14 shutdown date, not a same-day cutoff
OpenAI's API deprecation page carries an October 14, 2026 notice for selected GPT-4 series models. The company says those models will be removed from the API on January 14, 2027. Deprecation is advance notice of a future shutdown, not a statement that these API models stopped working on October 14.
The notice lists two GPT-4o snapshots: gpt-4o-2024-08-06 and gpt-4o-2024-11-20. It also lists gpt-4o-mini-2024-07-18, gpt-4.1-2025-04-14 and gpt-4.1-mini-2025-04-14. Naming the exact model identifiers matters when checking an application or vendor integration; a general GPT-4 label is not a complete inventory of the affected endpoints.
OpenAI names gpt-6-luna as the recommended replacement for each row in this notice. A replacement recommendation is not a promise that an existing application will behave identically after a model change. Teams should test their own prompts, response formats, tools and evaluations before moving a production workflow.
The page states that fine-tuned versions follow the same shutdown schedule as their underlying base models. That means a fine-tuned deployment is not exempt merely because it has a separate model identifier. The October 14 notice does not provide a separate extension window for these fine-tunes.
This announcement is specifically about the API models in its table. It does not establish a same-day retirement of GPT-5.5 across ChatGPT or Codex, or a new default for every subscription tier. Those product surfaces and their access rules need their own current documentation rather than an inference from an API shutdown table.
For buyers and administrators, the immediate work is to identify whether an application actually uses one of the listed snapshots, assess its dependencies and plan migration before January 14. The notice does not quote a migration price or guarantee a drop-in conversion. Keep the scheduled shutdown separate from today's availability, and confirm the replacement's current terms and performance for the specific workflow.