> ## Documentation Index
> Fetch the complete documentation index at: https://narev.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Benchmark OpenAI models in Narev with production data

> Connect OpenAI to Narev and benchmark GPT-4, GPT-4o, and other OpenAI models against your live production prompts before shipping changes.

<Tip>Test alternative OpenAI model configurations before deploying to production.</Tip>

## Setup options

<AccordionGroup>
  <Accordion title="Option 1: Use Narev gateway with OpenAI models">
    Call the Narev API directly and prefix any OpenAI model with `openai:`. Example: `model: 'openai:gpt-4o'`.
  </Accordion>

  <Accordion title="Option 2: Import from a tracing platform">
    If you use a tracing platform with your OpenAI integration, import traces from
    [Weights and Biases Weave](/platform/benchmark/integration/wandb),
    [LangSmith](/platform/benchmark/integration/langsmith), or
    [Langfuse](/platform/benchmark/integration/langfuse).
  </Accordion>

  <Accordion title="Option 3: Manual file upload">
    If you store OpenAI request and response history in files, create a benchmark, go to **Setup**, choose **Upload file**
    in **Data source**, upload your file, and save the project. For formats, see
    [File upload](/platform/benchmark/data-source/file-upload).
  </Accordion>

  <Accordion title="Option 4: Bring your own key (Enterprise)">
    Use your own OpenAI API key directly in Narev. This option is available on the enterprise tier.
  </Accordion>
</AccordionGroup>