🔍 Read the full analysis: OpenAI Model Migration: What To Expect As GPT-4o, GPT-4.1, And GPT-5-nano Switch Off on ThorstenMeyerAI.com
Get business pricing on monitors, keyboards and dev gear
- Business-only prices and quantity discounts
- Tax-exempt purchasing
- Multiple users, one account, clear invoices
TL;DR
OpenAI’s published API schedule sets Nov. 1, 2026 as a point when some listed model IDs have already stopped working, with more GPT-5 models due to retire Dec. 11 and GPT-4o and GPT-4.1 snapshots due Jan. 14, 2027. The listed replacements differ in price and use reasoning, so developers should test their applications rather than assume a model-name change will preserve behavior.
OpenAI has set API retirement dates for several GPT model families, with GPT-5, GPT-5 mini and GPT-5-nano scheduled to leave on December 11, 2026, followed by listed GPT-4o and GPT-4.1 snapshots on January 14, 2027. The changes affect developers whose applications call those models directly; OpenAI’s published schedule also names replacements that may differ in price and response behavior.
According to OpenAI’s deprecations page, checked by ThorstenMeyerAI.com on November 1, some older API model IDs had already been shut down on October 23, 2026. The listed models include gpt-4.1-nano, the original gpt-4o-2024-05-13 snapshot, gpt-4-turbo, gpt-3.5-turbo, o1 and o4-mini. Applications still sending requests to those IDs may fail, although the source does not provide a count of affected customers or failures.
The next major date is December 11. The schedule says gpt-5, gpt-5-mini and gpt-5-nano will be replaced by models in the gpt-5.6 family; o3, gpt-5-pro and o3-pro are also listed for retirement, with replacement options specified by OpenAI. On January 14, 2027, the scheduled retirements include dated snapshots of GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini. OpenAI lists gpt-6-luna as the replacement for those models.
The source reports that fine-tuned versions are scheduled to retire with their base models and that the listed replacements do not support fine-tuning. This could require more than changing a model ID: teams using custom fine-tunes may need to revise their application or ask OpenAI about continued access. The source also says OpenAI’s Evals platform, Agent Builder and reusable prompt objects are scheduled to shut down on November 30, while existing evals became read-only October 31.
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 Could Change Costs and Outputs
For businesses, a retirement can turn a model identifier buried in application code, a test or an automation into an availability risk. Teams need to find each use of the affected IDs and verify that requests still work after moving to a replacement. A successful API call alone may not mean the new model produces results in the format or style the application expects.
Price changes vary by migration path. ThorstenMeyerAI.com’s comparison of OpenAI’s listed short-context prices says GPT-4o’s input price is $2.50 per million tokens and its output price is $10.00; the listed GPT-6 Luna replacement is $0.10 input and $0.50 output. By contrast, the source says moving from GPT-5-nano to GPT-5.6 Luna raises the listed input price from $0.05 to $0.20 per million tokens and output from $0.40 to $1.20. These figures are a price comparison, not a forecast of an individual customer’s bill, which depends on usage and applicable pricing.
The source describes the replacements as reasoning models, which can use hidden reasoning tokens billed as output. It warns that certain sampling parameters may need to be removed and that a restrictive output limit could result in an empty response that still incurs a charge. Those are practical issues to check in a controlled test before switching production traffic.
Popular right now
KitchenAid Artisan 5-Quart Tilt-Head Stand Mixer, KSM150PS, Almond Cream
Vitamix 5200 Professional-Grade Blender, 64 oz. Container, Black
KitchenAid Artisan 5-Qt Tilt-Head Stand Mixer – Dried Rose (KSM150PS)
As an affiliate, we earn on qualifying purchases.
API Retirements Follow ChatGPT Changes
This schedule concerns OpenAI’s API, used by third-party applications and internal company systems. It is distinct from a change to the ChatGPT interface. The source says OpenAI removed GPT-4o, GPT-4.1 and GPT-4.1 mini from ChatGPT on February 13, 2026; that earlier removal does not itself mean the API models were shut down on the same date.
The deadlines apply to model IDs and snapshots listed on OpenAI’s deprecation page. The source says plain names such as gpt-4o and gpt-4.1 are marked deprecated in model documentation, while dated snapshots appear in the January schedule. Developers should check the exact identifier used in each request, rather than assuming that a short name and a dated snapshot have separate timelines.
The source also points to an October 29 revision of OpenAI’s Model Spec, the public document describing intended model behavior. It says the revision removed much of the detailed style guidance, including a default preference for Markdown, and asks models to follow the intent of instructions. The Spec is guidance about behavior, not itself a notice that an API model has been retired.
Notice Period and Compatibility Questions
OpenAI’s schedule gives October 14 as the notification date for the January 14 retirements, a period of about three months. The source says OpenAI’s policy promises at least six months’ notice for generally available models unless safety or compliance concerns call for a faster timeline. The cited notice does not explain which rule applies to these retirements or why the interval is shorter. The source suggests customers with contracts or committed spend ask their account teams; it also says some customers may be able to purchase dedicated capacity for continued access.
It is also unclear from the supplied material whether the replacements will preserve each customer’s application-level performance, response formats or quality on specific tasks. The listed prices do not establish total cost for a particular workload, and the source does not provide independent benchmark results. The new Model Spec’s behavior guidance likewise does not establish exactly how a given application’s outputs will change.
Test Before Each Retirement Date
Developers should first inventory model IDs across source code, configuration files, test fixtures, scheduled jobs and no-code automations. They should check those identifiers against OpenAI’s live deprecation page, since the source’s schedule was checked on November 1, 2026 and may change. Teams relying on models already listed as retired should investigate failing requests now rather than wait for the later deadlines.
Before the December and January changes, teams can run representative workloads against the named replacements and compare output quality, formatting, latency, error handling and token use. Requests that depend on a particular structure or style should state that requirement directly, then be tested for compliance. Organizations using fine-tuned models, evaluation tools or reusable prompts should check the separate shutdown dates and confirm what migration options are available.
The next scheduled milestones in the supplied material are the November 30 shutdown of Evals, Agent Builder and reusable prompt objects, the December 11 GPT-5-family retirements, and the January 14 GPT-4o and GPT-4.1 retirements. OpenAI’s deprecation page and account communications are the sources to monitor for changes, exceptions or dedicated-capacity arrangements.
Key Questions
When do the GPT-4o and GPT-4.1 API snapshots retire?
OpenAI’s schedule, as reported by ThorstenMeyerAI.com, lists specified GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini snapshots for retirement on January 14, 2027. Check the current deprecation page for the exact model IDs.
When are GPT-5, GPT-5 mini and GPT-5-nano scheduled to leave the API?
The listed retirement date is December 11, 2026. The source names GPT-5.6-family models as replacements, but developers should verify the current schedule and their own model IDs.
Will changing the model name preserve my application’s behavior?
That is not established by the supplied material. Because the replacements are described as reasoning models and may handle formatting, parameters and output limits differently, developers should test representative requests and specify required output formats.
Will the replacement models cost less?
It depends on the model being replaced. The source’s listed short-context prices show lower token prices for some GPT-4o-to-GPT-6 Luna comparisons, but higher prices for GPT-5-nano-to-GPT-5.6 Luna. Actual spending depends on usage and the applicable prices at the time.
What should teams do if they use fine-tuned models?
The source says fine-tuned models retire with their base models and that the listed replacements do not support fine-tuning. Teams should confirm their options with OpenAI and test an alternative before the relevant retirement date.
Source: ThorstenMeyerAI.com
