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

# Privacy and Logging

> Summary of how we handle prompts, outputs, and operational logs.

How NagaAI handles your prompts and outputs depends on whether you use paid or free models, and on whether you opt in to training for paid-model traffic.

In this page, **Service Data** means the prompts, inputs, and outputs sent through the service.

## At a glance

| Model type           | Training use                                                                                                |
| -------------------- | ----------------------------------------------------------------------------------------------------------- |
| Paid models, default | Service Data is used to serve the request and is not used for training by default                           |
| Paid models, opt-in  | If you enable **Training on paid models**, paid-model Service Data may be used for training and improvement |
| Free models          | Service Data may be collected and used by NagaAI and partners for training and improvement                  |

Specific provider practices can vary by model and provider.

## Paid models

When **Training on paid models** is disabled, paid-model Service Data is used to process the request and operate the service, not as training data.

If you enable **Training on paid models** in your dashboard privacy settings, that choice applies to future paid-model requests after you save it.

## Partners and model providers

Some requests are served through third-party model providers. For paid models, training use is tied to the paid-model training setting described above.

## Operational and security logs

To run and secure the service, NagaAI keeps limited operational and security logs such as:

* Rate limiting and abuse prevention signals
* Error diagnostics and reliability metrics
* Timestamps and high-level device/browser information
* Network identifiers (e.g., IP-related data)

These logs are used to operate, secure, and troubleshoot the service. They are separate from the training-use rules above.

## Aggregated usage data

NagaAI may collect anonymous, aggregated usage information such as overall model popularity to improve the platform. This data is not intended to identify you.

## Your choices

* Open your dashboard privacy settings to enable or disable **Training on paid models**.
* Changes apply to future requests after you save the setting.

For full legal terms, see the privacy policy below.

<Info>
  **Full Policy** For the complete details, refer to our [Privacy
  Policy](https://naga.ac/legal/privacy).
</Info>