A team's Foundry-hosted assistant has run on a specific model version for a year. Microsoft announces that version will be retired in three months, with a newer version available in the catalog. Before switching, what should the team do?
Select an answer to reveal the explanation.
Short Explanation
When a model version you depend on is scheduled to be retired, stand the replacement up next to what you already have and run your usual test questions through both, side by side, so you can see exactly where the new one behaves differently before anyone's real conversations depend on it. Waiting until the old version is actually gone just turns this into a fire drill, tearing down the working deployment and starting over throws away a known-good setup and skips the comparison, and turning up a capacity setting like throughput won't delay retirement at all.
Full Explanation
The correct answer is deploying the new model version alongside the existing one, running the team's established evaluation set against both, and comparing results before cutting over. This lets the team compare quality, safety, and behavior differences before any production traffic depends on the new version, catching regressions while the old version is still available as a fallback. Waiting until the old version is removed, then switching and monitoring for problems afterward is incorrect because it forces an unplanned, reactive cutover with no chance to compare behavior beforehand, risking an outage or a quality drop discovered only after users are already affected. Immediately deleting the current deployment and rebuilding from scratch is incorrect because it discards a working, tested configuration and skips any side-by-side comparison, so problems in the new version would only surface after the old one is already gone. Increasing the deployment's provisioned throughput to delay the retirement date is incorrect because provisioned throughput controls capacity and latency guarantees for a deployment; it has no bearing on a model version's retirement timeline, so increasing it does nothing to delay the deprecation deadline Microsoft has already announced.