When there are issues with API keys and provider settings, the user just sees "this is probably temporary" and doesn't get any feedback that they should check their API key. I think there are a few other situations where this comes up. e.g., I think this happens if we have an experiment that is tied to a LLM then you edit the provider and remove the model.
When there are issues with API keys and provider settings, the user just sees "this is probably temporary" and doesn't get any feedback that they should check their API key. I think there are a few other situations where this comes up. e.g., I think this happens if we have an experiment that is tied to a LLM then you edit the provider and remove the model.