Migration Guides
AI Stats migration guides cover two different upgrade problems:
- moving from one platform or gateway to AI Stats
- moving from an older model release to a newer model on the same platform
1. Model migrations
Use model migrations when the integration stays broadly the same but the model changes underneath it.
These guides focus on:
- request parameter changes
- prompt and reasoning behavior shifts
- structured output and tool-calling differences
- rollout, fallback, and rollback strategy
Current model migration guides:
Use platform migrations when you are changing SDKs, providers, or gateways and need to map your current integration onto AI Stats.
Current platform migration guides:
How to use these guides
- Identify whether you are changing the platform, the model, or both.
- Start with the guide that matches the riskiest part of the migration.
- Update request shape before comparing output quality.
- Run production-like evals before shifting live traffic.
Last modified on March 11, 2026
