Model unavailable errors across LLM providers

How model unavailable errors show up across Gemini, OpenAI, Anthropic, OpenRouter and more, and how to route around them.

Model-unavailable errors spike around deprecations, renames, and staged regional rollouts: the model id your agent was built with simply stops resolving. Hardcoded model names are the root cause — routing by capability instead of by id means a retired model never takes your agent down with it.