LLM Providers
=============

This application uses `OpenRouter`_ as the unified API gateway to access various Large Language
Models from different providers. OpenRouter provides a single interface to access models from
multiple LLM providers, simplifying integration and offering competitive pricing.

.. _OpenRouter: https://openrouter.ai/

The application supports several models from different providers, each optimized for different
use cases.


.. admonition:: About data policy

   The application restricts the list of providers from OpenRouter so that your data are never used
   to train or improve models.

   For some models though, the providers still retain data for a limited period of time (e.g. log
   retention to scan for abuse or for legal reasons). In the list of models below, see those that
   have a **zero data retention (ZDR)** policy: the providers serving these models will not store
   your data for any period of time.


Google Models
~~~~~~~~~~~~~

`Google`_ provides the Gemini family of models through OpenRouter:

* `gemini-2.5-pro <https://openrouter.ai/google/gemini-2.5-pro>`_: High-quality Gemini model for
  complex reasoning and analysis; excellent for interpretation tasks. *Zero data retention*
* `gemini-2.5-flash <https://openrouter.ai/google/gemini-2.5-flash>`_: Fast and efficient Gemini
  model; good balance of speed and quality for most tasks. *Zero data retention*

**Data Policy**: For Google Gemini models, this application uses only providers with a
**zero data retention policy**.

.. _Google: https://gemini.google.com/

OpenAI Models
~~~~~~~~~~~~~

`OpenAI`_ provides the GPT family of models through OpenRouter:

* `gpt-5 <https://openrouter.ai/openai/gpt-5>`_: Highest quality and most advanced model; best for
  complex interpretation and summarization, but more resource-intensive and slower than mini/nano.
* `gpt-5-mini <https://openrouter.ai/openai/gpt-5-mini>`_: Optimized for speed and efficiency;
  suitable for most workflows with a balance of quality and cost.
* `gpt-5-nano <https://openrouter.ai/openai/gpt-5-nano>`_: Fastest and lowest cost for simple
  extraction or tagging; not for detailed interpretations.

**Data Policy**: OpenAI stores data for **30 days** for abuse detection purposes only. Your data is
**not used to train** or improve models.

.. _OpenAI: https://openai.com/

Mistral Models
~~~~~~~~~~~~~~

`Mistral AI <https://mistral.ai/>`_ provides the Mistral family of models through OpenRouter:

* `mistral-small-3.2 <https://openrouter.ai/mistralai/mistral-small-3.2-24b-instruct>`_: Efficient
  Mistral model for balanced performance; good for various text processing tasks. *Zero data retention*

**Data Policy**: For Mistral AI models, this application uses only providers with a
**zero data retention policy**.

Custom Model Configuration
~~~~~~~~~~~~~~~~~~~~~~~~~~

You can configure the application to use your own models from different providers or custom endpoints.
To do so, go to ``Files`` > ``Preferences`` > ``LLM`` settings in Geoscience ANALYST.

OpenAI API compatible models
----------------------------

* **URL**: The OpenAI-compatible API endpoint from your model provider (e.g. ``https://yourmodelprovider.com/api/v1/``)
* **Model**: The name or identifier of the model you intend to use (e.g. ``your-model-name``)
* **Enter LLM API key**: Click to enter the API key for your model provider

Ollama models (local)
---------------------
* **Model**: The name of the Ollama model you have installed locally (e.g. ``mistral:7b``)

.. important::
   We cannot guarantee compatibility for custom models, as they might not fully support some API options
   and their answer format might differ from the expected one.