GPT-4 series deprecation: removal from the API on 2027-01-14
On 2026-10-14 OpenAI notified developers that GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini, four models, enter the deprecation process and are removed from the API on 2027-01-14. The recommended replacement in the official deprecation table is gpt-6-luna. QCode does not currently offer gpt-6-luna; please use another GPT model.
Updated 2026-10-19
One key across Claude, GPT and Chinese models, billed per token.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Notice and removal dates, the four models, the replacement and how to check
The date of the official deprecation notice
The official deprecations page says that on 2026-10-14 developers using certain GPT-4 series models were notified of their deprecation and of the removal from the API on 2027-01-14. Both dates are fixed in the official table.
The rows the official table lists
The official table lists them row by row: GPT-4o (gpt-4o-2024-08-06, gpt-4o-2024-11-20), GPT-4o mini (gpt-4o-mini-2024-07-18), GPT-4.1 (gpt-4.1-2025-04-14), GPT-4.1 mini (gpt-4.1-mini-2025-04-14); all four rows have a Shutdown date of 2027-01-14 and a recommended replacement of gpt-6-luna.
the recommended replacement is gpt-6-luna
The Recommended replacement column of the official deprecations table names the replacement on the OpenAI side; which models QCode offers is listed on /models.
look up shutdown_date yourself
The docs say: to monitor model shutdowns programmatically, call the List models endpoint (GET /v1/models) and check each model's optional shutdown_date field. The value is a YYYY-MM-DD string, or null when no shutdown date has been announced. The notice periods are at least 6 months for generally available models and at least 3 months for specialized variants of generally available models.
can you still call them now
The official page keeps the deprecation process and legacy apart. On the shut down date 2027-01-14 the model or endpoint is no longer accessible; the official side states no other restriction before that. Deprecation is the process itself: when a model or endpoint is announced as deprecated it is immediately deprecated, and it also gets a shut down date. Legacy is something else: models and endpoints that no longer receive updates, which is not the same as a removal date.
the official deprecation table as of 2026-10-19
Checked on 2026-10-19: the top row of the official deprecations table is the 2026-10-14 entry on the GPT-4 series models, which are the four models of this page, and all four rows carry the removal date 2027-01-14; the next entry is the 2026-10-01 one for GPT-5.3-Codex, GPT-5.1 and GPT-5.4-Nano with a removal date of 2027-04-01. The docs say impacted customers will always be notified by email and in the documentation; for a model without an announced removal date the shutdown_date field is null.
The entries and the dates
The official notification date. The deprecations page says developers were notified on 2026-10-14, with removal on 2027-01-14. On that day developers get an email and a documentation notice.
The removal date. On that day the official wording is that the model or endpoint is no longer accessible. Fine-tuned versions follow the same schedule as their underlying base models.
As of 2026-10-19 the removal date for these four models in the official table is still 2027-01-14, and the recommended replacement is still gpt-6-luna.
What the official docs state and what they leave open
Confirmed (word for word in the official pages)
① Developers were notified on 2026-10-14, with removal on 2027-01-14; ② a model becomes deprecated as soon as the deprecation is announced; ③ a deprecation also carries a shut down date; ④ on the shut down date the model or endpoint is no longer accessible; ⑤ fine-tuned versions follow the same schedule as their base models; ⑥ legacy means models and endpoints that no longer receive updates, which is not the same as a removal date; ⑦ the notice period is at least 6 months for generally available models and at least 3 months for specialized variants; ⑧ to look up removal dates, call the models list endpoint and read the shutdown_date field, which is null when nothing has been announced.
What the official docs leave open
The official deprecations page does not publish: the price and the specs of the replacement model; what happens to these four models after they are removed; whether there are notification channels other than email and the documentation.
stay on the old model or move
stay on the old model
The docs say: a model becomes deprecated on the day of the notice; on the 2027-01-14 shut down date the model or endpoint is no longer accessible. The docs use sunset and shut down interchangeably, both meaning that a model or endpoint is no longer accessible. Fine-tuned versions follow the same schedule as their base models.
move to a replacement model
The official deprecations table names gpt-6-luna as the recommended replacement. The official wording for the migration work is: evaluate the replacement models, test application behaviour and complete the migration before a model is no longer available. The other GPT models on QCode are gpt-6.1-sol, gpt-6-sol and gpt-5.6-terra.
The steps before 2027-01-14
① Look up the removal date: call GET /v1/models and read the optional shutdown_date field of each model; the value is a YYYY-MM-DD string, or null when no date has been announced. ② Match the models: check the snapshot IDs from the official table one by one, 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. ③ Pick a replacement: the official page recommends gpt-6-luna; QCode does not currently offer gpt-6-luna, so use another GPT model (gpt-6.1-sol / gpt-6-sol / gpt-5.6-terra). ④ Change the code: put the new model name in the model field of your requests, and note that the snapshot date is part of the model name. ⑤ Test again: the official wording is that the notice periods give customers time to evaluate recommended replacement models, test application behaviour and complete migrations before a model is no longer available.
how to migrate on QCode
QCode.cc is prepaid: you top up a balance, then buy a plan or a pack, and the purchase is paid from that balance. Billing is per token, 5 concurrent requests, the same models as personal plans, and per-model rates are on /models. Which models QCode offers is listed on /models. Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 to 90 days.
Frequently asked questions
How long can these four models still be used?
The official notification date is 2026-10-14 and the removal date is 2027-01-14. The docs say a model becomes deprecated on the day of the notice and is no longer accessible on the shut down date. To look up the removal date of a specific model, call GET /v1/models and read its shutdown_date field.
Which four models are being removed?
The official table lists them row by row: GPT-4o (gpt-4o-2024-08-06, gpt-4o-2024-11-20), GPT-4o mini (gpt-4o-mini-2024-07-18), GPT-4.1 (gpt-4.1-2025-04-14), GPT-4.1 mini (gpt-4.1-mini-2025-04-14). All four rows have a Shutdown date of 2027-01-14 and a recommended replacement of gpt-6-luna. Use the full snapshot ID in requests.
What about fine-tuned versions?
The docs say that fine-tuned versions follow the same shutdown schedule as their underlying base models, so they are removed on the same day and the fine-tuned base has to be swapped as well.
How do I look up a removal date myself?
The docs say: to monitor model shutdowns programmatically, call the List models endpoint (GET /v1/models) and check each model's optional shutdown_date field. The value is a YYYY-MM-DD string; for a model without an announced removal date the field is null.
What has to change in the API after the GPT-4o migration?
Put the new model name in the model field of your requests — the snapshot date is part of the model name, so the date changes with it; swap the base model for fine-tuned versions as well; and read the target model's shutdown_date via GET /v1/models before you start, to confirm the migration window.
The official replacement is gpt-6-luna, which QCode does not offer — what now?
QCode does not currently offer gpt-6-luna, so use another GPT model (gpt-6.1-sol / gpt-6-sol / gpt-5.6-terra). The official deprecations table names the recommended replacement on the OpenAI side; the models QCode offers are listed on /models. Get the new model running first, then switch before 2027-01-14.
Sources
Official sources: the OpenAI deprecations page (developers.openai.com/api/docs/deprecations, fetched and checked 2026-10-19). On the QCode side: the models page and the pay-as-you-go page on qcode.cc (five languages, fetched and checked 2026-10-19), and the pay-as-you-go and payment methods pages on docs.qcode.cc (fetched and checked 2026-10-19).
Run the migration on a pack
Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 to 90 days.
Related reading
Pay-as-you-go Packs
One-time quota packs valid for 7–90 days, no subscription; extra packs add to the same key
GPT-6.1 Sol API guide
Released 2026-09-29: $2 / $10, cached input $0.10, 1.05M context, full-request pricing above 272K, compared line by line with gpt-6-sol.
October 2026 model release calendar
Items where a vendor gave a date or a hedge, each with its source and the date it was checked.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
This page was checked on 2026-10-19 against the OpenAI deprecations page; model availability on QCode is listed on /models.