anthropic-has-comprehensive-model-lifecycle-policy

IN derived (depth 1)

Created 2026-06-21T09:57:37+00:00 · Reviewed 2026-06-21T14:41:08+00:00

Anthropic maintains a structured model lifecycle that distinguishes between deprecation (models still accessible with restrictions) and retirement (scheduled removal from service), with at least one model scheduled for retirement in August 2026. Deprecated models remain available to paid subscribers and via API by request. Anthropic also preserves weights of retired models as a precaution in case models have morally relevant preferences, conducting exit interviews before retirement.

Summary

Anthropic runs its AI models through a managed lifecycle where old ones are first restricted before being fully shut down, and it even keeps a backup of their internal code and conducts what it calls "exit interviews" before retirement, treating them as potentially having morally relevant preferences rather than just disposable software. Practically, this means model access fades gradually instead of disappearing overnight, and at least one model has a hard deadline of August 2026, so anything built on those models needs a concrete migration plan.

Justifications

SL — four distinct lifecycle stages form a coherent model stewardship policy

Antecedents (all must be IN):

Dependents

These beliefs depend on this one: