VertexAI Evals - Phoenix

Using Google Vertex AI with Phoenix Evals#

Info

Requires google-genai

pip install "arize-phoenix-evals>=3" google-genai

Create an LLM instance with the Google provider:

from phoenix.evals import LLMllm = LLM(provider="google", model="gemini-2.0-flash")

The LLM wrapper reads your API key from the GOOGLE_API_KEY environment variable, or you can authenticate via Application Default Credentials for Vertex AI.

Using with evaluators#

from phoenix.evals import LLM, evaluate_dataframefrom phoenix.evals.metrics import FaithfulnessEvaluatorllm = LLM(provider="google", model="gemini-2.0-flash")evaluator = FaithfulnessEvaluator(llm=llm)results_df = evaluate_dataframe(dataframe=df, evaluators=[evaluator])