🔍 Read the full analysis: GPT-4o, GPT-4.1, And GPT-5-nano Are Switching Off In The API—Now What? on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
OpenAI’s deprecation schedule lists API shutdowns for several GPT-5 models on Dec. 11, 2026, followed by GPT-4o and GPT-4.1 snapshots on Jan. 14, 2027. The source lists replacement models, but some may cost more and behave differently because they use reasoning; developers should check current OpenAI documentation and test their applications before migrating.
OpenAI is retiring API model IDs on a rolling schedule, with GPT-5, GPT-5 mini and GPT-5-nano set to leave on December 11, 2026, and listed GPT-4o and GPT-4.1 snapshots on January 14, 2027. The change affects developers and organizations whose apps call those models, which may need model-ID updates and testing before the deadlines.
According to the OpenAI deprecation schedule reviewed by ThorstenMeyerAI.com on November 1, models due to shut down on December 11 include GPT-5, GPT-5 mini, GPT-5-nano, GPT-5-pro, o3 and o3-pro. The source lists GPT-5.6 models as replacements, including GPT-5.6 Sol for GPT-5 and o3, GPT-5.6 Terra for GPT-5 mini, and GPT-5.6 Luna for GPT-5-nano. It says the pro models map to GPT-5.6 Sol with its pro reasoning mode.
On January 14, 2027, the schedule lists dated snapshots of GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini. ThorstenMeyerAI.com reports that OpenAI recommends GPT-6 Luna as their replacement. The article says some older model IDs, including GPT-4.1-nano, the original GPT-4o snapshot, GPT-4-turbo, GPT-3.5-turbo, o1 and o4-mini, were already scheduled to stop working on October 23. Calls to shut-down IDs may fail.
The source also reports that fine-tuned models follow their base model’s schedule, and that the listed replacements do not support fine-tuning. It identifies a separate November 30 closure affecting OpenAI’s Evals platform, Agent Builder and reusable prompt objects; existing evals became read-only on October 31, according to the article. Developers should verify these dates and model mappings against OpenAI’s current documentation before acting.
Infographic 01 · The shutdown calendar
What stops working, and when
OpenAI's API deprecations, as listed on 1 November 2026. GPT-4o and GPT-4.1 already left ChatGPT on 13 February 2026; this round is about apps built on the API.
- 23 Oct 2026Already offAlready shut down
gpt-3.5-turbo,gpt-4,gpt-4-turbo,gpt-4.1-nano,gpt-4o-2024-05-13,o1,o1-pro,o3-mini,o4-mini,gpt-image-1. Calls to these fail today. - 30 Nov 2026Your testsEvals platform, Agent Builder, reusable prompts
Not models, but tools many teams test with. Existing evals became read-only on 31 Oct. Export them first.
- 11 Dec 2026Notified 11 JunGPT-5 family and o3
gpt-5→gpt-5.6-sol·gpt-5-mini→gpt-5.6-terra·gpt-5-nano→gpt-5.6-luna·o3→gpt-5.6-sol·gpt-5-pro,o3-pro→gpt-5.6-sol(pro mode) - 14 Jan 2027Notified 14 OctGPT-4o, GPT-4o mini, GPT-4.1, GPT-4.1 mini
Snapshots
gpt-4o-2024-08-06,gpt-4o-2024-11-20,gpt-4o-mini-2024-07-18,gpt-4.1-2025-04-14,gpt-4.1-mini-2025-04-14→gpt-6-luna. Fine-tuned versions follow their base model. - Jan to Apr 2027Plan aheadAudio, speech, older GPT-5.x
6 Jan:
tts-1and others · 20 Jan: older audio and realtime models · 26 Feb:whisper-1, GPT-4o transcription · 1 Apr:gpt-5.1,gpt-5.4-nano,gpt-5.3-codex
Sources: OpenAI deprecations page (checked 1 Nov 2026) · OpenAI Help Center, model release notes. Dates as listed by OpenAI; shutdowns can be rescheduled.
Powered by Thorsten Meyer AI
Migration Costs and Behavior Changes
The schedule matters because switching a model is not necessarily a drop-in code change. The source describes the listed replacements as reasoning models, which can handle prompts differently and may require changes to sampling parameters. It also says hidden reasoning tokens are billed as output tokens, and that a restrictive output-token limit can produce an empty answer while still incurring a charge. Those details could affect application behavior and operating costs.
Pricing may change in either direction. ThorstenMeyerAI.com’s comparison of short-context standard rates says GPT-6 Luna’s listed input price is $0.10 per million tokens, compared with $2.50 for GPT-4o. But the source says GPT-5-nano users moving to GPT-5.6 Luna would face an input price four times higher, while GPT-5 mini users moving to GPT-5.6 Terra would pay eight times the input price. These are reported price comparisons, not a forecast of any individual customer’s bill; usage patterns and current prices need checking.
A model change can also affect output format and style. The article says OpenAI’s October 29 Model Spec revision removed much of its detailed style guidance, including a default to format answers in Markdown. For products that rely on a specific structure, developers may need to state formatting instructions directly and test whether the replacement follows them consistently.
Products Worth Considering
Apple MacBook Pro 14.2 with M5 Pro Chip, 18-Core CPU / 20-Core GPU, 64GB RAM, 1TB SSD, Silver
Apple MacBook Air 13-inch with M5 chip, 16GB memory, 1TB SSD, and 12MP Center Stage camera (Midnight)
Apple MacBook Pro with M5 Max 18-core CPU 20-core GPU 14.2-inch Display 48GB Memory 2TB SSD Space Black
As an affiliate, we earn on qualifying purchases.
The API Deprecation Timeline
This is an API change, not a new ChatGPT shutdown announcement. ThorstenMeyerAI.com says GPT-4o, GPT-4.1 and GPT-4.1 mini were removed from ChatGPT on February 13, 2026; the schedule at issue concerns models used through OpenAI’s developer API and the software built on it.
The source says OpenAI documents deprecations on its public schedule and emails affected customers. It also reports that the schedule offers at least six months’ notice for generally available models, unless safety or compliance concerns call for a faster timetable. The article notes that the January 14 notice was published October 14, a three-month interval, but says the schedule does not explain which notice rule applies. It reports that some customers may be able to buy dedicated capacity to keep access after a shutdown.
The article also describes an October 29 revision to OpenAI’s public Model Spec, the stated guide to model behavior. According to its account, the revision removed many detailed style provisions and instructs models to follow the intent of instructions. It quotes OpenAI as saying its models “do not yet fully reflect” the Spec. That statement concerns the Spec and model behavior; it does not establish how any particular replacement will perform in a given application.
Open Questions Before Migration
The schedule and pricing details here are reported from a source article that says it checked OpenAI’s pages on November 1, 2026. Prices and documentation can change, so the exact current rates, model availability and replacement mappings should be confirmed directly with OpenAI. The source does not establish whether all affected customers received notices or whether every listed replacement will meet a particular application’s needs.
It is also unclear why the reported notice for the January 14 shutdown is three months despite the policy language described in the article. The article says the schedule does not identify whether a safety or compliance exception applies. The extent of any dedicated-capacity access, its terms and availability are also not specified. Finally, the source provides no independent benchmark results showing how the replacement models compare on accuracy, latency or reliability for individual workloads.
Tests to Run Before Shutdowns
Developers should first inventory every API model ID used in application code, configuration files, test fixtures and no-code workflows, then match each ID to the current official deprecation schedule. The October 23 date means the first step for any still-used retired ID is to check whether requests are already failing. Teams with contracts or committed spend can ask their OpenAI account contact about timing and continued-access options.
Before December and January, teams can run the proposed replacement against representative production prompts in a test environment. They should check output quality and format, latency, token usage, reasoning-token charges, sampling settings, and behavior under output limits. Any model-specific fine-tuning or dependence on Evals, Agent Builder or reusable prompt objects also needs a migration plan in light of the separate November 30 dates reported by the source. OpenAI’s current documentation and customer notices will determine the applicable details.
Key Questions
When are GPT-5 and GPT-4o API models scheduled to shut down?
The schedule reported by ThorstenMeyerAI.com lists GPT-5 and several related models for December 11, 2026, and listed GPT-4o and GPT-4.1 snapshots for January 14, 2027. Check OpenAI’s live deprecation page for the current model list and dates.
Does this schedule mean ChatGPT is shutting down these models now?
No. The development covered here concerns API model IDs used by developers’ applications. The source separately says GPT-4o, GPT-4.1 and GPT-4.1 mini had already been removed from ChatGPT on February 13, 2026.
Will the replacements cost less?
Not in every case. The source reports lower listed input and output prices for some moves to GPT-6 Luna, but higher rates for GPT-5-nano users moving to GPT-5.6 Luna and GPT-5 mini users moving to GPT-5.6 Terra. Verify current rates and compare them with your actual token usage.
What should developers test before switching?
Test representative prompts and workflows for quality, formatting, latency, token use, reasoning charges and output-limit behavior. Also check sampling parameters, fine-tuning dependencies and any planned retirement of developer tools used in your workflow.
Are all shutdown details settled?
No. The source says the reason for the reported three-month notice before the January shutdown is not explained. Current prices, customer-specific notices, continued-access terms and replacement performance for individual applications should be confirmed with OpenAI.
Source: ThorstenMeyerAI.com
