The list · Being kept

A retired model stays open to the public after its retirement date

Retirement normally means a model can no longer be used by anyone outside the company that made it. A developer can instead keep a retired model available. This entry tracks whether that has happened, and on what terms.

Political progress First pass

Where it stands · offered today

Happened once, for one model, at one developer

A page Anthropic published on February 25th, 2026 says one of its retired models remains open to paying users of its chat product, and to others on request. The developer says it is not committing to do the same for every model. Elsewhere it has gone the other way: on January 29th, 2026 OpenAI announced that on February 13th, 2026 it would withdraw from its chat product an older model that it had earlier restored after users objected. It said nothing would change for now in the service through which developers use its models.

Offered by Anthropic. As of November 17th, 2026. Source: Anthropic's update on the retirement of its model Claude Opus 3 (Anthropic, February 25th, 2026).

What would count

At least one leading AI model that its developer lists as retired can still be used by members of the public, by the developer's own published account.

What settles it

The developer's own dated page, together with its retirement page listing the model as retired.

What Is on Record About AI Models, By Kind

kept apart, never added together

No AI model has yet been asked any of the questions this record will keep, because the written method for asking does not exist yet. Nothing is recorded below for this entry. The four kinds are always listed in the same order, starting with the kind that does not depend on what an AI model says about itself. None of them settles whether an AI model wants anything, in either direction. Why the questions have not been asked yet.

  1. Measured without the AI model's own word

    Nothing recorded.

  2. Reports checked against an independent reading

    Nothing recorded.

  3. Choices made at a cost

    Nothing recorded.

  4. Plain statements

    Nothing recorded.

Three Cautions

recorded with the same care as the rest

If an AI system has no wants, and is treated as if it had

The head of Microsoft AI cites the steps taken for this model as an example of treating models "as though they have feelings, preferences, rights, or any entitlement to our welfare", and argues that doing so will make advanced AI harder to keep under control. Source: An essay by the head of Microsoft AI warning against work on model welfare (Mustafa Suleyman, September 16th, 2026).

If an AI system has wants, and they go unnoticed

Nothing recorded yet from anyone but the developer. The developer's own reason is given beside its page under Sources, below.

If an AI system has wants, and they put people at risk

The developer's page frames continued access partly as lowering a safety risk. Source: Anthropic's update on the retirement of its model Claude Opus 3 (Anthropic, February 25th, 2026).

What Could Not Be Confirmed

looked for, and not settled
  • Whether the retired model is still available today. The developer's page says so as of February 25th, 2026.
  • Whether the essays the retired model was given a place to publish are still appearing after the three months stated.
  • A safety researcher's argument about this step taken on its own. A search found none.
  • Whether the other developer's withdrawal took place on February 13th, 2026. The page opened announces it in advance.
  • How the retired model is open to developers on request, when the developer's list of retired models says that requests to retired models fail. Both pages say what they say, and both may be true.

Sources

each one opened on the date given
  1. Anthropic's update on the retirement of its model Claude Opus 3 Anthropic, February 25th, 2026. Opened November 19th, 2026. The developer’s own account. Says the retired model remains available to all paying users of the chat product and by request to developers, and that it was given a place to publish essays for three months at the least. It counts possible risks to the welfare of the models themselves among the downsides of retiring a model, and says that recording a model's stated preferences, and acting on them where that costs little, is worthwhile for the models and for those who use them.
  2. Anthropic's list of the models it has retired or will retire Anthropic. Opened November 17th, 2026. No date on the page. The developer’s own account. Lists that model as retired on January 5th, 2026, and says that requests to retired models fail.
  3. OpenAI's notice that it is withdrawing GPT-4o and older models from its chat product OpenAI, January 29th, 2026. Opened November 17th, 2026. The developer’s own account. Says OpenAI will withdraw GPT-4o from its chat product on February 13th, 2026, having earlier restored it after feedback from users who "preferred GPT-4o's conversational style and warmth". The reasons given concern users only.
  4. An essay by the head of Microsoft AI warning against work on model welfare Mustafa Suleyman, September 16th, 2026. Opened November 17th, 2026.

This entry is a first pass, written from the sources above and added on November 17th, 2026. A second check compared it with those sources on November 17th, 2026. Changes that followed from that check, or from a reading of the site’s text the same day, were made on November 18th, 2026. Those changes were checked on November 18th, 2026. More was added to this entry on November 19th, 2026, when every entry was given a third caution. That addition was checked on November 19th, 2026. What has been checked.