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

# AI Providers

> Set-up AI API keys

This guide providers more details about creating and setting AI api keys in your .env file for using Xyne's chat feature.

To use **AI chat**, you need to configure access credentials. You can use one of the following providers: ***AWS, OpenAI, Ollama, Gemini, Together AI or Fireworks AI***.

### AWS (Amazon Bedrock)

* Sign in to the [AWS Management Console](https://console.aws.amazon.com/console/home)
* Generate your `Access Key` and `Secret Key`.
* Add them as environment variables in your `.env.default` file:
  ```javascript .env.default theme={null}
  AWS_ACCESS_KEY=<YOUR_AWS_ACCESS_KEY> 
  AWS_SECRET_KEY=<YOUR_AWS_SECRET_KEY>  
  AWS_REGION=<YOUR_AWS_REGION>     
  ```
  <Note> The `AWS_REGION` key is optional in the `.env.default` file; if not provided, it defaults to `us-west-2` </Note>

### OpenAI

* Create an API key from the [OpenAI API settings](https://platform.openai.com/settings/organization/api-keys).
* Add it as environment variables in your .env.default file:
  ```javascript .env.default theme={null}
  OPENAI_API_KEY=<YOUR_OPENAI_API_KEY> 
  ```

### Ollama (Local LLM)

* Install Ollama and the desired model from [Ollama](https://ollama.com/).
* Add the model name in your .env.default file:
  ```javascript .env.default theme={null}
  OLLAMA_MODEL=<YOUR_OLLAMA_MODEL_NAME> 
  ```

<Tip> If you're using Ollama, ensure the specified model is downloaded locally. </Tip>
<Tip> Also ensure that your downloaded ollama model is working in the host machine, you can use `ollama run your_model_name your_message`. This should return an output confirming ollama is working.</Tip>

<Warning> For `quickstart` mode ensure Ollama is listening from all addresses, so that docker is able to reach it. </Warning>

### Google AI

* Generate your API key from [Google AI Studio](https://aistudio.google.com/app/apikey)
* Add them as environment variables in your `.env.default` file:
  ```javascript .env.default theme={null}
  GEMINI_API_KEY=<YOUR_GEMINI_API_KEY> 
  GEMINI_MODEL=<YOUR_GEMINI_MODEL_NAME>  
  ```

### Together AI

* Create your API key in [Together AI](https://www.together.ai/)
* Select the model from your choice.
* Add the following to your .env.default file:
  ```javascript .env.default theme={null}
  TOGETHER_API_KEY=<YOUR_TOGETHER_API_KEY>
  TOGETHER_MODEL=<YOUR_TOGETHER_MODEL>
  TOGETHER_FAST_MODEL=<YOUR_TOGETHER_FAST_MODEL>
  ```

### Fireworks AI

* Create your API key in [Fireworks AI](https://fireworks.ai/)
* Select the model from your choice.
* Add the following to your .env.default file:
  ```javascript .env.default theme={null}
  FIREWORKS_API_KEY=<YOUR_FIREWORKS_API_KEY>
  FIREWORKS_MODEL=<YOUR_FIREWORKS_MODEL>
  FIREWORKS_FAST_MODEL=<YOUR_FIREWORKS_FAST_MODEL>
  ```

## Setting up for Reasoning \[Together | Fireworks | Ollama] :

For addition of only reasoning in your model, you need to add REASONING in your .env.default file:

```javascript .env.default theme={null}
REASONING=true
```

### For example :

#### Reasoning with Together AI :

```javascript .env.default theme={null}
TOGETHER_API_KEY=<YOUR_TOGETHER_API_KEY>
TOGETHER_MODEL=deepseek-ai/DeepSeek-R1
TOGETHER_FAST_MODEL=deepseek-ai/DeepSeek-V3
REASONING=true
FAST_MODEL_REASONING=false
```

#### Reasoning with Fireworks AI :

```javascript .env.default theme={null}
FIREWORKS_API_KEY=<YOUR_FIREWORKS_API_KEY>
FIREWORKS_MODEL=accounts/fireworks/models/deepseek-r1
FIREWORKS_FAST_MODEL=accounts/fireworks/models/deepseek-v3
REASONING=true
FAST_MODEL_REASONING=false
```

#### Reasoning with Ollama :

```javascript .env.default theme={null}
OLLAMA_MODEL=<REASONING_MODEL>
OLLAMA_FAST_MODEL=non reasoning model
REASONING=true
FAST_MODEL_REASONING=false
```

#### Additional BASE URL support

We've also added support for adding `base url` values for OpenAI and Together AI, along with your Open AI or Together API Key, you can the `base url` value:

```javascript theme={null}
OPENAI_API_KEY=<YOUR_OPEN_API_KEY>  # If you want to use OpenAI
    ### OR
TOGETHER_API_KEY=<YOUR_TOGETHER_API_KEY>  # If you want to use Together AI
BASE_URL=<YOUR_BASE_URL>
```

### Key Selection Order :

If multiple keys are set in the environment variable, our system follows this precedence:

```javascript theme={null}
AWS → OpenAI → Ollama → Together AI → Fireworks AI → Gemini
```

<Tip>Only a single provider is required to get started. This means either AWS or OpenAI, Ollama, Gemini, Together AI or Fireworks AI can be used, all of them aren't required.</Tip>