Skip to main content

How Search Party Queries Models

An overview of how search party gathers prompt responses.

Q
Written by Quinn Schwartz
Updated over 3 weeks ago

How It Works

Search Party connects directly with all major large language models as an API partner. The default method is querying models via API, which delivers the most consistent and reliable responses.

In addition to API calls, Search Party also simulates web browser experiences for certain models. To capture geographic variation, the system proxies requests through specific locations, allowing you to see how responses differ based on local context.

This blended approach—API-first, supplemented with browser simulation and geographic proxies—ensures accuracy while reflecting the real-world variety of model outputs.

Key Benefits

  • Accuracy: API queries provide clean, direct access to model outputs

  • Realism: Browser simulation mirrors how end users might see results

  • Geographic Insight: Proxying by location shows local differences in responses

  • Balance: Multimodal approach gives both stability and flexibility

Common Use Cases

  • Marketing: Comparing how brand or product queries show up across regions

  • Product marketing: Testing feature perception in different geographies

  • Global teams: Understanding local visibility in priority markets

Step-by-Step Instructions

  1. Add prompts in the Prompts section

  2. By default, Search Party queries models via API

  3. For models requiring simulation, the system runs a browser-based query

  4. To capture local differences, enable geographic proxying

  5. Review outputs in analytics to see both consistent API responses and local variations

Did this answer your question?