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

# Models We Use

> The AI models behind Exterview's agents and Smaya: frontier and open-source providers, added over time, with your data not used to train shared models.

## Purpose

Exterview's evaluation agents and Smaya run on a mix of frontier and open-source models. We're building the ability to pick the best model for each task based on quality, latency, and cost, and we're adding providers over time. This page names the providers in use today and the guarantees that hold regardless of which model runs.

<Note>
  This list is current as of 5 July 2026. Providers can change as models evolve.
  Contact your Exterview representative for the latest set.
</Note>

## Providers we work with

We're adding these providers over time; not all of them are fully wired up yet. Contact your Exterview representative to confirm which ones are live for your setup today.

<CardGroup cols={2}>
  <Card title="Anthropic (Claude)">
    Reasoning and evaluation across agent tasks.
  </Card>

  <Card title="OpenAI">Reasoning and generation across agent tasks.</Card>

  <Card title="Google (Gemini)">Reasoning and multimodal tasks.</Card>

  <Card title="Mistral">Efficient models for targeted tasks.</Card>

  <Card title="Microsoft">
    Azure-hosted models, aligned with the Azure-native platform.
  </Card>

  <Card title="Open-source models">
    Selected open-weight models for specific tasks.
  </Card>
</CardGroup>

## The guarantees that do not change

Whichever model runs a given task, the same controls apply.

<CardGroup cols={2}>
  <Card title="No training on your data">
    We don't use your candidate and configuration data to train shared models.
  </Card>

  <Card title="Processed under your organization's boundary">
    Content is processed under your organization's boundary, under Exterview's
    data-protection terms.
  </Card>

  <Card title="Content-based, not biometric">
    Scoring is grounded in what a candidate says, not facial analysis, tone, or
    other biometrics.
  </Card>

  <Card title="Model choice is ours to route">
    Exterview is building the ability to pick the best AI model for each task, and is adding providers over time; you rely on the outcome and its guarantees, not a specific vendor.
  </Card>
</CardGroup>

## FAQs

<AccordionGroup>
  <Accordion title="Do you train models on our data?">
    No. We don't use your candidate and configuration data to train shared models.
  </Accordion>

  <Accordion title="Can we pin a specific model or provider?">
    Model routing is managed by Exterview to keep quality and cost balanced.
    Specific requirements can be discussed with your representative.
  </Accordion>

  <Accordion title="Where does model processing happen?">
    Under your organization's boundary, under Exterview's data-protection terms. Providers that process data appear on the [Sub-processors](/trust/sub-processors) page.
  </Accordion>
</AccordionGroup>

## Related

<CardGroup cols={2}>
  <Card title="Sub-processors" href="/trust/sub-processors">
    The third parties that process data, by purpose.
  </Card>

  <Card title="Privacy & Residency" href="/trust/privacy-residency">
    Consent, retention, deletion, and residency.
  </Card>

  <Card title="Responsible AI" href="/trust/responsible-ai">
    Human review and no autonomous adverse decisions.
  </Card>

  <Card title="AI Observability" href="/monitor/ai-observability">
    How model outputs are evaluated and audited.
  </Card>
</CardGroup>