GPT-5.1, gpt-5.3-codex, and gpt-5.4-nano shutdown on April 1, 2027
The dataset has 5 OpenAI ids with shutdown date 2027-04-01: gpt-5.1, gpt-5.1-2025-11-13, gpt-5.3-codex, gpt-5.4-nano, gpt-5.4-nano-2026-03-17. OpenAI's deprecations page says: "The following models are deprecated and will be removed from the API on April 1, 2027, with six months' notice. Migrate to the recommended replacements before the shutdown date." The same page names gpt-6-sol for gpt-5.1, gpt-5.1-2025-11-13, and gpt-5.3-codex. It names gpt-6-luna for gpt-5.4-nano and gpt-5.4-nano-2026-03-17. The section heading is 2026-10-01. OpenAI's deprecations page also says: "At the time of the shut down, the model or endpoint will no longer be accessible."
Last reviewed . Publisher: SunsetDiff.
Steps
Read the five ids. The table lists gpt-5.1, gpt-5.1-2025-11-13, gpt-5.3-codex, gpt-5.4-nano, and gpt-5.4-nano-2026-03-17. gpt-5.1-chat-latest, gpt-5.1-codex, gpt-5.1-codex-max, and gpt-5.1-codex-mini are a different date.
Swap the model string. Send gpt-6-sol in place of gpt-5.1, gpt-5.1-2025-11-13, and gpt-5.3-codex. Send gpt-6-luna in place of gpt-5.4-nano and gpt-5.4-nano-2026-03-17. Keep a gpt-5.3-codex call on the Responses API.
Check effort, tools, and sampling fields. gpt-5.1 and gpt-5.4-nano default reasoning effort to none. gpt-6-sol and gpt-6-luna default it to medium. On Chat Completions, function calling for those two replacements works with reasoning_effort set to none. When effort is not none, the Using GPT-6 page says to remove temperature, top_p, and top_logprobs. Test one real prompt before April 1, 2027.
Models shutting down
These rows are the OpenAI ids in the dataset with shutdown date 2027-04-01. The model id links to its SunsetDiff model page. Deprecation date, shutdown date, and replacement are the fields stored on that row. Not published means the field is empty.
OpenAI's deprecations page says: "The following models are deprecated and will be removed from the API on April 1, 2027, with six months' notice. Migrate to the recommended replacements before the shutdown date." The section heading is 2026-10-01. These rows store announced_date 2026-10-01. deprecation_date is empty. The same page says: "When we announce that a model or endpoint is being deprecated, it immediately becomes deprecated. All deprecated models and endpoints will also have a shut down date."
The shutdown table lists gpt-5.3-codex with replacement gpt-6-sol. It lists gpt-5.4-nano and snapshot gpt-5.4-nano-2026-03-17 with replacement gpt-6-luna. It lists gpt-5.1 and snapshot gpt-5.1-2025-11-13 with replacement gpt-6-sol. The plain-text alias list at the bottom of that page pairs gpt-5.4-nano with gpt-5.4-nano-2026-03-17, and gpt-5.1 with gpt-5.1-2025-11-13. Both strings in each pair shut down on 2027-04-01.
The gpt-5.1 model page says the default snapshot is gpt-5.1-2025-11-13. The gpt-5.4-nano model page says the default snapshot is gpt-5.4-nano-2026-03-17. The gpt-5.3-codex model page says the default snapshot is gpt-5.3-codex. That page does not list a second dated snapshot.
gpt-5.1-chat-latest, gpt-5.1-codex, gpt-5.1-codex-max, and gpt-5.1-codex-mini are not in the table above. OpenAI's deprecations page says: "On April 22, 2026, we announced the deprecation of the following older OpenAI models. Access to these models was shut down on July 23, 2026." The July 23, 2026 table lists gpt-5.1-chat-latest, gpt-5.1-codex, and gpt-5.1-codex-max with substitute gpt-5.6-sol. It lists gpt-5.1-codex-mini with substitute gpt-5.6-terra. The stored shutdown date on those four rows is 2026-07-23.
On April 1, 2027 these ids reach the shutdown date stored in the dataset. OpenAI's deprecations page says: "At the time of the shut down, the model or endpoint will no longer be accessible." The same page uses sunset and shut down for that moment. It does not print an HTTP status for it.
A 400 from a field the replacement rejects is a different failure. Those sentences are on the request errors pages. This guide does not invent an error body for the shutdown itself.
Search the repository for the alias and for the dated snapshot. gpt-5.1 and gpt-5.1-2025-11-13 can sit in different files. The same is true for gpt-5.4-nano and gpt-5.4-nano-2026-03-17. How to find every model ID in your codebase covers SDK argument names. The scan page builds a search from dataset ids that have a shutdown date.
OpenAI's deprecations page says to call the List models endpoint, GET /v1/models, and read each model's optional shutdown_date field. When present, the field is a YYYY-MM-DD string, or null if no shutdown date has been announced. The same field is on the Retrieve a model endpoint.
You can also read one id from the free SunsetDiff lookup file. gpt-5.1 is one of these rows. The API page lists the other files. No key and no signup. The alerts page is the free email form. That page says what the form stores and the one welcome email it can send.
Send gpt-6-sol where the code now sends gpt-5.1, gpt-5.1-2025-11-13, or gpt-5.3-codex. Send gpt-6-luna where the code now sends gpt-5.4-nano or gpt-5.4-nano-2026-03-17. The deprecations page names those replacements. It does not name gpt-6.1-sol for this shutdown.
The gpt-5.1 model page lists Chat Completions (v1/chat/completions), Responses (v1/responses), and Batch (v1/batch) as supported. The gpt-5.4-nano model page lists the same three routes as supported. The gpt-5.3-codex model page lists Responses as supported. It lists Chat Completions and Batch as not supported. Keep a gpt-5.3-codex call on Responses.
The gpt-6-sol model page lists Chat Completions, Responses, and Batch as supported. The gpt-6-luna model page lists the same three routes as supported. Both pages list fine-tuning (v1/fine-tuning) as not supported. The gpt-5.1, gpt-5.3-codex, and gpt-5.4-nano pages also list fine-tuning as not supported.
gpt-5.1 and gpt-5.1-2025-11-13 on Chat Completions
The gpt-5.1 model page says: "Reasoning.effort supports: none (default), low, medium, and high." The gpt-5.4-nano model page says: "Reasoning.effort supports: none (default), low, medium, high and xhigh." The gpt-5.3-codex model page says: "GPT-5.3-Codex supports low, medium, high, and xhigh reasoning effort settings." That page does not name none, max, or a default.
The gpt-6-sol model page and the gpt-6-luna model page both say: "reasoning.effort supports none, low, medium (default), high, xhigh, and max." A gpt-5.1 or gpt-5.4-nano request that leaves effort unset uses none. The same omitted field on gpt-6-sol or gpt-6-luna uses medium. Set the effort you want. The Using GPT-6 page says: "Use reasoning.effort in Responses or reasoning_effort in Chat Completions." It also says: "GPT-6 Astra and GPT-6.1 Sol do not support the none reasoning effort; GPT-6.1 Terra, GPT-6 Sol, and GPT-6 Luna do."
The gpt-6-sol model page and the gpt-6-luna model page both say: "Use the Responses API for built-in tools and function calling. Chat Completions supports function calling only with reasoning_effort set to none." The Using GPT-6 page says: "GPT-6 Sol and GPT-6 Luna support function calling in Chat Completions only with reasoning_effort: "none". Use Responses for reasoning with tools." A Chat Completions call that sends function tools sets reasoning_effort to none, or the call moves to the Responses API. The recorded error for gpt-6-luna is on Function tools with reasoning_effort are not supported for gpt-6-luna. The gpt-6-sol model page states the same rule. This guide does not invent an error sentence for gpt-6-sol.
The Using GPT-6 page says: "When reasoning effort is not none, remove temperature, top_p, and top_logprobs. For Chat Completions, also remove logprobs. For Responses, remove message.output_text.logprobs from include." The page for that temperature sentence is temperature is not supported with this model. gpt-6-sol and gpt-6-luna can keep temperature when reasoning effort is none.
The gpt-5.1 model page lists a 400,000 context window and 128,000 max output tokens. The gpt-5.3-codex and gpt-5.4-nano pages list a 400,000 context window, maximum input tokens 272,000, and 128,000 max output tokens. The gpt-6-sol and gpt-6-luna pages list a 1,050,000 context window, maximum input tokens 922,000, and 128,000 max output tokens.
The numbers below are the text token prices on each model page, per 1M tokens. The gpt-5.1 page covers the alias and lists snapshot gpt-5.1-2025-11-13. The gpt-5.4-nano page covers the alias and lists snapshot gpt-5.4-nano-2026-03-17. Those snapshot ids do not have a second price table on those pages.
The gpt-6-sol page and the gpt-6-luna page both say: "Prompts with more than 272K input tokens are priced at 2x input and cache rates and 1.5x output for the full request." The gpt-5.4-nano page says: "Regional processing (data residency) endpoints are charged a 10% uplift for GPT-5.4 nano." The gpt-6-sol page and the gpt-6-luna page both say: "Regional processing adds a 10% premium where available."
Text token prices per 1M tokens from the model pages. Not listed means that page has no cache writes row.
OpenAI's deprecations page says the notice period is time to evaluate the recommended replacement, test application behavior, and finish the migration before the model is no longer available. Do this before April 1, 2027.
The pre-migration test checklist is the longer list: evals, rejected fields, token limits, tools, price, and latency.
Search the repository for gpt-5.1, gpt-5.1-2025-11-13, gpt-5.3-codex, gpt-5.4-nano, and gpt-5.4-nano-2026-03-17.
Replace gpt-5.1, gpt-5.1-2025-11-13, and gpt-5.3-codex with gpt-6-sol.
Replace gpt-5.4-nano and gpt-5.4-nano-2026-03-17 with gpt-6-luna.
Keep a gpt-5.3-codex call on the Responses API. That model page lists Chat Completions as not supported.
Set reasoning effort on the replacement. none on gpt-5.1 and gpt-5.4-nano becomes medium on gpt-6-sol and gpt-6-luna when the field is left unset.
On Chat Completions with function tools, set reasoning_effort to none, or move the call to the Responses API.
When reasoning effort is not none, remove temperature, top_p, and top_logprobs. On Chat Completions, also remove logprobs. On Responses, remove message.output_text.logprobs from include.
Send one prompt you use now to the replacement id before April 1, 2027.
Call GET /v1/models and read shutdown_date for each id you still send.
The dataset lists gpt-5.1, gpt-5.1-2025-11-13, gpt-5.3-codex, gpt-5.4-nano, gpt-5.4-nano-2026-03-17 with shutdown date 2027-04-01. The deprecations page says: "The following models are deprecated and will be removed from the API on April 1, 2027, with six months' notice. Migrate to the recommended replacements before the shutdown date."
What is the recommended replacement?
gpt-5.1, gpt-5.1-2025-11-13, and gpt-5.3-codex use gpt-6-sol. gpt-5.4-nano and gpt-5.4-nano-2026-03-17 use gpt-6-luna. The recommended replacement is the official replacement stored on each row. gpt-5.1: gpt-6-sol. gpt-5.1-2025-11-13: gpt-6-sol. gpt-5.3-codex: gpt-6-sol. gpt-5.4-nano: gpt-6-luna. gpt-5.4-nano-2026-03-17: gpt-6-luna.
Do gpt-5.1-chat-latest and the older Codex ids shut down on the same day?
Those ids are gpt-5.1-chat-latest, gpt-5.1-codex, gpt-5.1-codex-max, and gpt-5.1-codex-mini. Their stored shutdown date is 2026-07-23. The deprecations page says: "On April 22, 2026, we announced the deprecation of the following older OpenAI models. Access to these models was shut down on July 23, 2026." The substitute for gpt-5.1-chat-latest, gpt-5.1-codex, and gpt-5.1-codex-max is gpt-5.6-sol. The substitute for gpt-5.1-codex-mini is gpt-5.6-terra.
Is a model string swap enough?
The string change is gpt-6-sol or gpt-6-luna, matching the row. The gpt-5.1 page says: "Reasoning.effort supports: none (default), low, medium, and high." The gpt-6-sol page says: "reasoning.effort supports none, low, medium (default), high, xhigh, and max." Leaving effort unset changes none to medium. The Using GPT-6 page says: "When reasoning effort is not none, remove temperature, top_p, and top_logprobs. For Chat Completions, also remove logprobs. For Responses, remove message.output_text.logprobs from include." A gpt-5.3-codex call stays on the Responses API. That model page lists Chat Completions as not supported.
What does OpenAI say happens at shut down?
OpenAI's deprecations page says: "At the time of the shut down, the model or endpoint will no longer be accessible." That page does not print an HTTP status for that moment.
Is a deprecation date stored?
No deprecation_date is stored on these rows. The deprecations section is dated 2026-10-01, and the dataset stores 2026-10-01 as announced_date. The page says: "When we announce that a model or endpoint is being deprecated, it immediately becomes deprecated. All deprecated models and endpoints will also have a shut down date."