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

# Supported LLM providers and models

> The LLM providers ABBYY Vantage supports for prompt-based extraction, including connection fields and example models for Anthropic, Google AI Studio (Gemini), Microsoft Foundry, Mistral AI, OpenAI, and any OpenAI-compatible endpoint you host yourself.

ABBYY Vantage integrates external LLM providers through **LLM connections**, using credentials you supply for that provider or endpoint. The following providers are supported.

## Supported providers

| Provider | Description | Connection fields | Example models |
| :- | :- | :- | :- |
| **Anthropic** | Direct connection to Anthropic's Claude models. | **Model Name**, **API Key**; optional **Upstream API Base** | `claude-fable-5`, `claude-opus-4-8`, `claude-sonnet-5`, `claude-haiku-4-5` |
| **Google AI Studio (Gemini)** | Google's Gemini models through AI Studio. | **Model Name**, **API Key**; optional **API Base** | `gemini-2.0-flash`, `gemini-2.5-pro` |
| **Microsoft Foundry** | OpenAI and other AI models hosted on Microsoft's Azure infrastructure. | **Model Name**, **API Base**, **Azure API Key** | `gpt-4`, `gpt-4.1-nano`, `gpt-35-turbo` |
| **Mistral AI** | Direct connection to Mistral AI's models. | **Model Name**, **API Key** | `mistral-large-latest`, `mistral-medium-latest`, `mistral-small-latest` |
| **OpenAI** | Direct connection to OpenAI's API. | **Model Name**, **OpenAI API Key**; optional **API Base** and **OpenAI Organization ID** | `gpt-5.6`, `gpt-4` |
| **Custom (OpenAI-Compatible)** | Any endpoint that implements the OpenAI chat completions API, including a model you host yourself. | **Model Name**, **API Base**, **API Key** | Whatever model your endpoint serves, for example `mistral-large-2411` |

The models listed are examples, not a complete list. Enter the exact model or deployment name that your provider account supports.

<Info>
  **ABBYY Phoenix Plus** is not listed above. It is an LLM endpoint ABBYY supplies and operates, so it has no connection fields to configure and its model is selected by ABBYY rather than by you. See [LLMs in ABBYY Vantage](/vantage/documentation/llms/llms).
</Info>

To set up a connection to any of these providers, see [Configure LLM connections](/vantage/documentation/skills/connections).

## Anthropic

Direct access to Anthropic's Claude models with a simple, API-key-only setup. Best for organizations that want to use Claude models for extraction and classification.

* Connection fields: **Model Name** and **API Key** (starts with `sk-`). An **Optional** section adds **Upstream API Base**.
* Document data is processed on Anthropic's infrastructure, subject to Anthropic's terms and data policies.

## Google AI Studio (Gemini)

Google's Gemini models through AI Studio, with a simple, API-key-only setup. Best for cost-effective access to multimodal models, especially for organizations already using Google Cloud.

* Connection fields: **Model Name** and **API Key**. An **Optional** section adds **API Base**.
* Document data is processed on Google's infrastructure, subject to Google's terms and data policies. Model availability may vary by region.

## Microsoft Foundry

OpenAI and other AI models hosted on Microsoft's Azure infrastructure. Best for enterprise deployments that need data-residency controls, Azure integration, or custom model hosting.

* Connection fields: **Model Name** (the exact name of your deployed model), **API Base** (the full endpoint, including the deployment path and API version), and **Azure API Key**.
* Available models may differ from OpenAI direct. Suited to organizations with an existing Azure subscription and compliance requirements.

## Mistral AI

Direct access to Mistral AI's models with a simple, API-key-only setup. Best for organizations that want Mistral models, including cost-effective options for high-volume extraction.

* Connection fields: **Model Name** and **API Key**.
* Document data is processed on Mistral AI's infrastructure, subject to Mistral AI's terms and data policies.

## OpenAI

Direct access to OpenAI models with a simple, API-key-only setup. Best for organizations already using the OpenAI platform.

* Connection fields: **Model Name** and **OpenAI API Key** (starts with `sk-`). An **Optional** section adds **API Base** and **OpenAI Organization ID**.
* Document data is processed on OpenAI's infrastructure, subject to OpenAI's terms and data policies.

## Custom (OpenAI-Compatible)

An endpoint you host or control that implements the OpenAI chat completions API. Best for a privately hosted model, or for a provider that is not in the list.

* Connection fields: **Model Name** (the name your endpoint serves), **API Base** (the endpoint's base URL, for example `https://my-llm.example.com/v1`), and **API Key**.
* The endpoint must be reachable from your ABBYY Vantage deployment over `http` or `https`. Vantage checks the URL format when you save the connection, and reachability only when you test it.
* Document data is processed wherever that endpoint runs, under whatever terms apply to it.

## Related topics

* [LLMs in ABBYY Vantage](/vantage/documentation/llms/llms)
* [Configure LLM connections](/vantage/documentation/skills/connections)
* [Extract data with prompt-based activities](/vantage/documentation/skills/prompt-extraction)