Google Vertex AI Chat (Cached)Install

Summary

Generate chat completions using Google Vertex AI's cached chat model, optionally leveraging a pre-existing cache to return responses efficiently or configure generation parameters for dynamic output.

Inputs

• Project ID (required) — Google Cloud project ID to use for the chat request; can be selected from a list or entered manually.
• Model Name (required) — The name of the Vertex AI chat model to use, e.g., 'gemini-2.5-flash'.
• Cached Content Name — Full resource name of the cached content to use; if provided, generation parameters are ignored and responses come from the cache.
• Options — Collection of optional settings including location (Google Cloud region), sampling parameters (temperature, topP, topK), maximum output tokens, thinking budget (controls reasoning token allocation), safety settings (content filtering by category and threshold), and Google Search grounding (enables grounding responses with live Google Search results; incompatible with cached content).

Output shape

a single AI language model chat response object

If a Cached Content Name is provided, the node returns cached completions instead of generating new ones. Enabling Google Search grounding cannot be used with cached content; the node warns accordingly and disables grounding in that case. Safety settings and other generation options apply only when no cached content is used.