Portkey

DeepEval's integration with Portkey AI allows you to use the portkey gateway to connect to any model to power all of DeepEval's metrics.

Command Line

To configure your Portkey model through the CLI, run the following command:

deepeval set-portkey \
    --model "your-model" \ # Ex: gpt-4.1
    --provider "your-provider" \ # Ex: openai
    --base-url "your-base-url" \
    --temperature=0

info

The CLI command above sets Portkey as the default provider for all metrics, unless overridden in Python code. To use a different default model provider, you must first unset Moonshot:

deepeval unset-portkey

Persisting settings

You can persist CLI settings with the optional --save flag. See Flags and Configs -> Persisting CLI settings.

Python

Alternatively, you can define PortkeyModel directly in python code:

from deepeval.models import PortkeyModel
from deepeval.metrics import AnswerRelevancyMetric

model = PortkeyModel(
    model="gpt-4.1",
    provider="openai",
    api_key="your-api-key",
    base_url="your-base-url"
)

answer_relevancy = AnswerRelevancyMetric(model=model)

There are ZERO mandatory and FIVE optional parameters when creating a PortkeyModel:

• [Optional] model: A string specifying the name of the Portkey model to use. Defaults to PORTKEY_MODEL_NAME if not passed; raises an error at runtime if unset.
• [Optional] api_key: A string specifying your Portkey API key for authentication. Defaults to PORTKEY_API_KEY if not passed; raises an error at runtime if unset.
• [Optional] provider: A string specifying the Portkey provider of your model. Defaults to PORTKEY_PROVIDER_NAME if not passed; raises an error at runtime if unset.
• [Optional] base_url: A string specifying the base URL for the model API. Defaults to PORTKEY_BASE_URL if not passed; raises an error at runtime if unset.
• [Optional] generation_kwargs: A dictionary of additional generation parameters forwarded to Portkey's completion(...) / acompletion(...) call

tip

Any **kwargs you would like to use for your model can be passed through the generation_kwargs parameter. However, we request you to double check the params supported by the model and your model provider in their official docs.