[Enhancement]: Ability to switch LLM models if an API encounters errors for an agent

[distardao]

What features would you like to see added?

A feature to set up a list of LLM APIs for an agent, allowing the agent to choose one or more of them under specific conditions.

More details

I noticed that the Anthropic API has been very unstable recently. Can we have a feature to configure LLM APIs in a specific order, by priority, or based on conditions to ensure agents always remain operational?

Which components are impacted by your request?

No response

Pictures

No response

Code of Conduct

• I agree to follow this project's Code of Conduct

[distardao]

I think this list of LLM APIs could come from different providers. This would allow the service provided by the agent to be more stable, which is very important for the user experience in any business.

[harrisonhxy]

I've encountered this question a lot in my use.
Although I'm looking forward to this feature, I'm considering whether the feature should be better implemented at the backend or API proxy level?