OpenAI
OpenAI retires GPT-4o and GPT-4.1 from the API on Jan. 14
OpenAI will remove GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini from the API on Jan. 14, 2027, and points all four to GPT-6 Luna.
OpenAI will remove GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini from its API on Jan. 14, 2027, and recommends GPT-6 Luna as the replacement for all four. Developers using them must move to another model before then.
OpenAI’s deprecations page, read Oct. 13, carries an entry dated Oct. 14 under the heading “GPT-4 series model deprecations.” The entry says OpenAI notified developers using certain GPT-4 series models of the removal.
Four Models and Their Fine-Tunes Shut Down Together
The shutdown covers five snapshots: gpt-4o-2024-08-06, gpt-4o-2024-11-20, gpt-4o-mini-2024-07-18, gpt-4.1-2025-04-14 and gpt-4.1-mini-2025-04-14. OpenAI said fine-tuned versions follow the same schedule as their base models.
GPT-4.1 nano is not in the deprecation table. OpenAI’s pricing page, read Oct. 13, still lists it at $0.10 per 1M input tokens and $0.40 per 1M output tokens.
Luna Costs Less Than Every Model It Replaces
Luna’s standard price is $0.10 per 1M input tokens and $0.50 per 1M output tokens, according to OpenAI’s model pages read Oct. 13. That is below all four retiring models on both counts. GPT-4o costs $2.50 and $10. GPT-4o mini costs $0.15 and $0.60. GPT-4.1 costs $2 and $8. GPT-4.1 mini costs $0.40 and $1.60.
Prompts over 272K input tokens are billed at 2x the input rate and 1.5x the output rate for the full request. Luna’s context window is 1,050,000 tokens with up to 128,000 output tokens. GPT-4o and GPT-4o mini stop at 128,000 and 16,384. GPT-4.1 and GPT-4.1 mini allow 1,047,576 and 32,768. Luna’s knowledge cutoff is May 18, 2026, against Oct. 1, 2023 for GPT-4o and June 1, 2024 for GPT-4.1.
Luna launched Sept. 22, 2026, according to the OpenAI API changelog.
Luna Reasons Before It Answers
Luna is a reasoning model. Its page lists a reasoning.effort setting that takes none, low, medium, high, xhigh or max. Medium is the default. OpenAI calls Luna “our most efficient model for focused, high-volume tasks.”
The page tells developers to use the Responses API for built-in tools and function calling. The Chat Completions API supports function calling on Luna only when reasoning effort is set to none.
Our View
The deprecation table lists dated snapshot IDs. The Jan. 14 date applies to those exact versions, such as gpt-4o-2024-11-20 or gpt-4.1-2025-04-14. Fine-tuned versions share the same date.
Luna reasons, at medium effort unless told otherwise. The Luna page limits Chat Completions function calling to effort none. A team that sends tool calls that way has to set effort to none or move to Responses. Either is a code change to finish before January.
The lower price helps only if the reasoning setting leaves output tokens where they were. Compare bills on Luna at the effort you plan to use before counting on the savings.