Is there any method to use a "fallback" provider in case the main one is gone?

[JusefPol]

Hi guys, I have been checking the docs and I don't see a way to provide a second provider and model in case the main one is having issues. Is there any plan to implement something like this? it would be great so that we can use a local model as main, but if failing we can use a different provider from cloud for example and the switch is automatic.

Thanks.

[nicoloboschi]

hey not at the moment. on our roadmap we have to add a llm gateway or (integrate an existing one) so you can do it a proxy level, not application level

[nicoloboschi]

it's implemented now https://hindsight.vectorize.io/developer/configuration#multi-llm-strategies-failover--round-robin