Which AI models are being retired, and when

Past their retirement date (84)

These ids no longer answer. If one is still in your code, or in something an assistant wrote for you, it is already failing in production.

Retiring within 90 days (39)

Dated, imminent, and enforced. This is the work queue.

Dated, further out (70)

A deadline exists but is not imminent. Worth knowing before you pin a new dependency on one of these.

Deprecated with no date yet (3)

Superseded and no longer recommended, but still served. A retirement date usually follows, so treat these as migrations you can schedule rather than emergencies.

By provider

Which of these are you actually paying for?

This board tells you what is retiring. It cannot tell you which of these your own traffic hits, or what the forced move costs. One line of code meters the models your app really calls, then puts a dollar figure on each retiring id and on the switch to its replacement, including when that switch costs more.

Prefer to consume this as data? The same records are available from the keyless API, as an RSS feed, or over MCP so your coding agent can check a model id before it ships one.