model deprecation

The process by which a model provider announces that a model version will be withdrawn from its model APImodel API, n.: A web service through which programs send requests to an AI model run by a company and receive its answers, usually paying per use instead of running the model themselves. or applications and then retires it on a set date, after which requests fail or are redirected to a successor. Providers expose dated snapshots that developers can pin to avoid silent behavior changes, alongside aliases that always point to the latest version, and deprecation schedules usually give months of notice. For engineering teams, a deprecation forces migration work: re-running evaluationsevaluation, n.: Testing an AI system to find out what it can and cannot do, how well, and how safely — from exam-style quizzes to hands-on trials and expert reviews. and golden setsgolden dataset, n.: A small, carefully checked set of example questions with known good answers, used as a trusted yardstick to test an AI system every time it changes., re-tuning prompts written around the old model's quirks, and checking changes in cost and latencylatency, n.: How long you wait between asking an AI system for something and getting the answer. Lower latency means a snappier, more responsive system.. Because hosted models can disappear, retirement also raises questions of research reproducibility, of users' attachment to a familiar model's personality and, for some researchers, of model welfareAI welfare, n.: Concern for the well-being of AI systems themselves, in case some of them can suffer or have interests that matter — and the research into how to tell and what to do about it.. For providers, retiring old models frees serving capacity and ends the work of maintaining safeguards on them.

The deprecation notice gave us six months to move the classifier off the old snapshot and re-run every eval.