When Your LLM Reaches End-of-Life: A Framework for Confident Model Migration in Production Systems

TL;DR

This paper introduces a Bayesian framework for confidently migrating production LLM systems by calibrating automated metrics with human judgments, demonstrated on a large-scale commercial QA system.

Contribution

It presents a novel Bayesian approach for model comparison that effectively combines automated metrics and limited human evaluations for enterprise LLM migration.

Findings

Successfully identified suitable replacement models for a large-scale QA system.

Calibrated automated evaluation metrics against human judgments using Bayesian methods.

Framework applicable broadly to enterprise LLM deployment and migration.

Abstract

We present a framework for migrating production Large Language Model (LLM) based systems when the underlying model reaches end-of-life or requires replacement. The key contribution is a Bayesian statistical approach that calibrates automated evaluation metrics against human judgments, enabling confident model comparison even with limited manual evaluation data. We demonstrate this framework on a commercial question-answering system serving 5.3M monthly interactions across six global regions; evaluating correctness, refusal behavior, and stylistic adherence to successfully identify suitable replacement models. The framework is broadly applicable to any enterprise deploying LLM-based products, providing a principled, reproducible methodology for model migration that balances quality assurance with evaluation efficiency. This is a capability increasingly essential as the LLM ecosystem continues to evolve rapidly and organizations manage portfolios of AI-powered services across multiple models, regions, and use cases.