🔍 Read the full analysis: GPT-4o, GPT-4.1, And GPT-5-nano Are Leaving OpenAI’s API: A Developer Guide on ThorstenMeyerAI.com
Get audio and creator gear 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 published deprecation schedule says several API models, including GPT-5-nano, GPT-4o and GPT-4.1 snapshots, will stop working on set dates. Replacement models may differ in price and behavior, and the source says developers should test prompts, settings and output limits before migrating. Some dates and pricing details should be checked against OpenAI’s live documentation.
OpenAI plans to remove several models from its API, including GPT-5-nano on December 11, 2026, and GPT-4o and GPT-4.1 snapshots on January 14, 2027, according to the deprecation schedule described by ThorstenMeyerAI.com. Developers using those model IDs will need to migrate or risk API calls failing after the relevant date; the proposed replacements can also change costs and response behavior.
The source says the December shutdown list includes GPT-5, GPT-5 mini, GPT-5-nano, GPT-5-pro, o3 and o3-pro. It identifies suggested replacements as 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. GPT-5-pro and o3-pro are listed for GPT-5.6 Sol with its pro reasoning mode.
On January 14, the listed models include dated snapshots of GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini. The article says OpenAI recommends GPT-6 Luna for those models. It also reports that several older IDs, including GPT-4.1-nano, GPT-4 Turbo, GPT-3.5 Turbo, o1 and o4-mini, were scheduled to stop working on October 23, 2026. Developers should confirm each ID’s current status in OpenAI’s documentation.
The price comparisons in the source use standard per-million-token rates for prompts up to 272,000 tokens, checked on November 1. It lists GPT-4o at $2.50 per million input tokens and $10 per million output tokens, versus $0.10 and $0.50 for GPT-6 Luna. By contrast, GPT-5-nano’s listed replacement, GPT-5.6 Luna, costs $0.20 per million input tokens and $1.20 per million output tokens, compared with $0.05 and $0.40 for GPT-5-nano. These are the source’s reported prices, not a guarantee of current rates.
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 Can Change Cost and Output
The schedule matters to teams whose applications, automations or tests call a retiring model by ID. Once a model is unavailable, requests using that identifier may fail, potentially interrupting a user-facing service or an internal workflow. A migration is not necessarily a like-for-like swap: the source’s examples show some replacements priced lower and others higher on the stated token rates.
The replacement models are described as reasoning models, which can affect both billing and implementation. The source says hidden reasoning tokens are billed as output, some sampling parameters must be removed, and a tight output-token limit can result in an empty answer while still incurring a charge. Teams should test representative tasks and monitor token use rather than relying only on the model name or a basic success response.
Behavior may change even when a request continues to run. The source reports that OpenAI revised its Model Spec on October 29, removing much of its detailed style guidance, including a default for Markdown formatting. Developers who require JSON, a particular structure or a consistent tone should specify those requirements in their prompts and test the resulting outputs.
Products Worth Considering
Samsung 27 Odyssey OLED G5 (G50SF) QHD QD-OLED Gaming Monitor
Samsung 27-inch Odyssey OLED G6 QD-OLED QHD Gaming Monitor
LG 34GX90SA-W 34-inch Ultragear WQHD OLED Curved Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Two API Changes on Separate Timelines
This development concerns OpenAI’s API, not only the models available in ChatGPT. The source says GPT-4o, GPT-4.1 and GPT-4.1 mini were removed from ChatGPT on February 13, 2026, but says the later dates concern API access and therefore services built by developers.
The source describes OpenAI’s deprecation page as the basis for the shutdown calendar and says it was checked on November 1, 2026. It reports that OpenAI emails affected customers and documents deprecations on that page. It also says fine-tuned versions follow the shutdown schedule of their base model, while the proposed replacement models listed in the article do not support fine-tuning.
A related tooling change is listed for November 30, 2026: the source says OpenAI plans to shut down its Evals platform, Agent Builder and reusable prompt objects at v1/prompts. It reports that existing evals became read-only on October 31. These dates are distinct from the model shutdowns and may affect how teams test or manage their migrations.
“If your app depends on a format, say so in the request.”
— ThorstenMeyerAI.com, summarizing the API migration risk
Dates, Exceptions and Live Pricing
The provided source says the January 14 retirement notices were issued on October 14, while OpenAI’s stated notice policy generally provides at least six months for generally available models unless safety or compliance concerns call for a faster schedule. The source says the deprecation entry does not explain which rule applies. It also does not establish whether any specific customer has a contractual exception or will receive continued access through dedicated capacity.
The schedule and prices are reported as checked on November 1, 2026, and may change. The material provided here does not independently verify the live OpenAI documentation, the scope of every affected model alias, or whether each listed replacement is available to every customer. Developers should check their account notices and current model and pricing pages before acting.
The source gives only partial details about migration behavior and does not include a complete, model-by-model compatibility table. It is therefore unclear which prompts, tools, sampling settings and fine-tuned workflows will transfer without adjustment. The effects of the Model Spec revision on any particular application also remain unconfirmed.
Audit IDs and Test Replacements
Developers should first search source code, configuration files, test fixtures and no-code workflows for every affected model ID, including dated snapshots and fine-tuned models based on retiring models. The next step is to confirm each shutdown date and proposed replacement in OpenAI’s current deprecation documentation, then check account-specific notices and pricing.
Before changing production traffic, teams can run representative evaluations against the replacement: compare answer quality and formatting, check tool calls and structured output, remove unsupported parameters, and test output limits and token costs. Where a model provides reasoning, budget for output-token billing and verify that the application handles empty or incomplete responses. The source also reports a November 30 deadline for certain OpenAI development tools, so teams that use Evals, Agent Builder or v1/prompts should review those dependencies separately.
The immediate milestones in the reported schedule are the November 30 tooling shutdown, the December 11 GPT-5 and o3 retirements, and the January 14 GPT-4o and GPT-4.1 retirements. OpenAI’s published documentation and customer notices will determine whether any of those dates, model mappings or prices change.
Key Questions
The source reports that GPT-5-nano, GPT-5, GPT-5 mini, GPT-5-pro, o3 and o3-pro are scheduled for removal on December 11, 2026. Confirm the date and replacement in OpenAI’s current deprecation documentation.
When are GPT-4o and GPT-4.1 API models scheduled to stop working?
The source lists dated GPT-4o, GPT-4o mini, GPT-4.1 and GPT-4.1 mini snapshots for January 14, 2027. It says GPT-6 Luna is the recommended replacement, but developers should check the exact model IDs and current guidance.
Will the replacements cost less?
Not in every case, according to the price comparisons in the source. It reports lower listed token rates for the GPT-4o-to-GPT-6 Luna move, but higher rates for GPT-5-nano moving to GPT-5.6 Luna and GPT-5 mini moving to GPT-5.6 Terra. Rates can change, so check the live pricing page.
Can developers keep fine-tuned versions of retiring models?
The source says fine-tuned models follow their base model’s shutdown schedule. It also says the listed replacements do not support fine-tuning, making it important for affected teams to confirm their available migration options with OpenAI.
What should developers test before switching?
Check prompts, formatting, tool use, sampling parameters, output limits and token costs. The source says the replacement models use reasoning tokens billed as output and that strict output limits can produce empty answers, so test these cases before routing production requests to a replacement.
Source: ThorstenMeyerAI.com
