OpenAI or Anthropic is deprecating a model we use in production
AI vendors retire models faster than most cloud services. A model that works perfectly today can be sunset in six to twelve months. Version pinning, an eval suite and a migration plan prevent your production from breaking suddenly.
Try this first
- 1Always pin a specific model version in code, not an alias like 'latest' or 'newest'. For example 'claude-sonnet-4-5' or 'gpt-4o-2024-08-06', not 'gpt-4'.
- 2Subscribe to the deprecation mailing or changelog of your vendors. Anthropic and OpenAI typically announce months ahead.
- 3Keep the eval suite ready. When a model is deprecated, run the suite against the successor. Within a day you know if migration is safe or needs a prompt tweak.
- 4Build a fallback path: if your primary model dies, route to a second provider or local model. For SMB this need not be a multi-provider gateway, a simple if-else with a second API key often suffices.
- 5Document which version runs where. On an incident you want to trace 'which model gave what at time X'.
When to bring us in
Running a production AI app without a migration protocol yet, we can set up the pattern and fallback for you.
See also
- Can I paste a customer file or email into ChatGPT?Depends on the account and settings. Free ChatGPT and a Team tenant behave very differently from what most people assume.
- I want a one-page AI policy for my teamA real one-pager beats a thick document nobody reads. Four headers and concrete examples.
- How do I tell if an AI answer is made up?Models sound confident even when they are wrong. A few habits catch most mistakes.
None of the above fits?
Describe your situation below. We pass your input plus the steps you already saw to our AI and return tailored next-step advice. If it's too risky to DIY, we'll say so.
Or skip the DIY entirely
Our Managed IT clients do not look these things up. One point of contact, a fixed monthly price, resolved within working hours.