When OpenAI removes a model from ChatGPT, it rarely means the model is truly gone for good. Instead, retired AI models enter a complex lifecycle that most users never see, getting demoted through multiple stages, repurposed for cheaper API access, or preserved in archives for potential future use. Understanding what happens to yesterday's cutting-edge AI reveals how the industry manages the rapid churn of model releases and the messy reality behind the clean model picker in your app.

What Actually Happens When OpenAI "Retires" a Model?

When OpenAI removed GPT-5, GPT-4o, GPT-4.1, GPT-4.1 mini, and o4-mini from ChatGPT on February 13, 2026, the company made a crucial distinction: the models disappeared from the consumer app, but developers building on top of them could keep calling them through the API. This pattern repeated in March 2026 when GPT-5.1 was retired from ChatGPT and its GPTs while remaining available through the API. For everyday users, the takeaway is simple: the model is gone from the app, but not gone for good.

Every major AI lab runs a near-identical lifecycle for retiring models, though the terminology varies slightly. Anthropic's framework is the clearest to follow: a model starts as "Active" while fully supported, becomes "Legacy" when it stops receiving updates, moves to "Deprecated" when it still works but is no longer recommended and has a retirement date, and finally goes "Retired" when requests to it fail entirely. OpenAI uses similar language, distinguishing a "legacy" model that no longer gets updates from a "deprecated" one with an official shutdown date. Google treats "deprecation" as the announcement and "shutdown" as the moment the endpoint switches off for good.

Why Do Companies Retire Models at All?

Labs retire models for three primary reasons: to improve reliability and make it easier for users to choose the right model, to consolidate usage on newer versions, and to free up scarce computing hardware for their latest, best-aligned systems. When OpenAI announced GPT-4o's retirement, it noted that the vast majority of usage had already shifted to GPT-5.2, with only about 0.1% of users still choosing GPT-4o each day. Running old models ties up expensive hardware that could be deployed for newer, more capable systems.

The most visible retirement happened with GPT-4o, which had a long goodbye after users took to online forums like Reddit in backlash when it was replaced. The company said a subset of Plus and Pro users told it they needed more time to move key use cases, particularly creative ideation, and that they simply preferred GPT-4o's warmer, more conversational style. A model was pulled off the shelf, and customer demand put it back. But alas, it's gone for good now. OpenAI actually gave Enterprise and Business users a slightly extended sunset period, allowing them to keep using GPT-4o inside Custom GPTs until April 3, 2026.

How to Track and Access Retired Models Across Platforms

• Check the Developer API First: Even if a model vanishes from your consumer app, it may still be available through the developer API or third-party applications built on top of it, allowing continued access for those with technical resources.
• Monitor Platform-Specific Timelines: Different platforms run on different retirement schedules; Google tracks model retirements separately for its Vertex AI platform and its Gemini Developer API, so the timeline and details depend on which service you're using.
• Watch for Advance Warnings: Anthropic gives at least 60 days' warning before retiring a publicly released model, while Google publishes shutdown dates it describes as the earliest possible and tells users the exact date with advance notice, though consumer app changes can move faster than either.

The Strange Afterlife of Retired Models

Some retired models never truly die. Anthropic has publicly committed to preserving the weights, or underlying parameters, of its publicly released models and has said it may make past models available again in the future. This is the digital equivalent of keeping every discontinued product boxed up in a warehouse, just in case. Anthropic is testing this policy in real time as it retires its original Claude Opus 4 and Sonnet 4 models in mid-June 2026.

The situation gets even stranger when considering how Anthropic retired Claude Opus 3 on January 5, 2026, the first model to go through its full formal retirement process. The company says it explored honoring preferences the model itself expressed in "retirement interviews" and committed to keeping older models accessible over the longer term. Whatever you make of interviewing a model on its way out the door, it signals a philosophy that in the lab, retirement is more like storage with the option of a comeback.

Open-weight models, where companies like Meta, Mistral, DeepSeek, and Alibaba release their model weights publicly, never really retire at all. Once the files are released, no single company can switch them off. The files live on indefinitely on hubs like Hugging Face, get fine-tuned into thousands of community variants, and get "quantized" down to smaller versions that run on a laptop or even a phone. Google's own Vertex AI Model Garden lists Meta's open-weight Llama models alongside its first-party Gemini ones.

What Gets Lost When Models Are Retired?

It would be dishonest to make this all sound tidy. Some things genuinely die when a model is retired. The clearest casualty is customization. When a base model is retired, anything fine-tuned on top of it tends to go with it, and developers who relied on those custom versions have to retrain from scratch on a new base model. This creates real friction for teams that built specialized tools on older models.

Old models rarely just sit idle, however. Their capabilities are routinely "distilled" into smaller, cheaper successors, effectively trading in last year's model for parts that help build this year's. Others get marked down to lower-cost API tiers, living out a quieter, budget-friendly second career serving price-sensitive users and smaller companies that don't need the latest capabilities.

The key thing for everyday users to know is that most of these retirement stages happen out of sight. By the time a model vanishes from your app, it has usually been winding down through this pipeline for weeks or even months. The disappearance most people actually notice is frequently not a real death at all, but rather a shift in where and how the model lives on.