Send a single-turn prompt to Gemini via Vertex AI — morie_vertex_ask

R port of morie.vertex.ask_gemini. POSTs to the Vertex AI REST endpoint :generateContent and returns the concatenated text from the first candidate.

Usage

morie_vertex_ask_gemini(
  prompt,
  model = NULL,
  system = NULL,
  temperature = 0.1,
  max_output_tokens = 2048L,
  timeout_s = 120,
  cfg = NULL
)

Arguments

prompt: Character scalar – the user prompt.
model: Optional Gemini model override.
system: Optional system instruction.
temperature: Numeric. Default 0.1.
max_output_tokens: Integer. Default 2048.
timeout_s: Numeric HTTP timeout. Default 120.
cfg: Pre-resolved config list, or NULL to auto-resolve.

Value

Character scalar – trimmed generated text.

Examples

if (FALSE) { # nzchar(Sys.getenv("GOOGLE_CLOUD_PROJECT")) || nzchar(Sys.getenv("MORIE_EE_PROJECT"))
# Runs only when a Google Cloud project is configured; Vertex is an
# OPTIONAL fallback -- the default LLM path is local Ollama.
morie_vertex_ask_gemini("Summarize the MRM framework in one line.")
}