Use "prompt_cache_retention" for OpenAI nodes

Feature Requests
, ,
1

The idea is:

Being able to use “prompt_cache_retention” property on OpenAI node.

My use case:

I want to have faster responses and pay less for AI usage.

I think it would be beneficial to add this because:

Faster and cheaper OpenAI node execution

Any resources to support this?

Are you willing to work on this?

I would like to, yes.

2

+1

Adding technical context: the main benefit of exposing prompt_cache_retention in the n8n OpenAI node (n8n-nodes-langchain.openai) is for workflows that repeatedly call the same system prompt with only dynamic input changes. Currently, the cache duration is fixed (24h), which may limit reuse efficiency for low-frequency executions or testing environments. Allowing configuration would enable better token usage, reduce unnecessary API calls, and optimize cost management, particularly in production scenarios with multiple clients or repeated orchestration patterns.