It is now possible to configure how often the LLM action will retry and when it will timeout. This is particularly helpful when using custom LLM providers that have inconsistent latency for responses.