Chat (LLM) Provider Options

Chat (LLM) providers include AI language models that generate responses to user queries based on the provided context.

Built-in Providers

Cohere

  • Availability: Built-in, always available

  • Name: cohere

  • Options:

    • model (required): The name of the Cohere language model to use.

    • timeout (default: 15): The request timeout in seconds.

    • temperature (default: 0): The sampling temperature for generating responses.

    • frequency_penalty (default: 0): The frequency penalty to apply to generated tokens.

    • presence_penalty (default: 0): The presence penalty to apply to generated tokens.

OpenAI

  • Availability: Built-in, always available

  • Name: openai

  • Options:

    • model (required): The name of the OpenAI language model to use.

    • timeout (default: 15): The request timeout in seconds.

    • max_tokens (default: 0): The maximum number of tokens to generate in the response.

    • organization_id (default: ''): The ID of your OpenAI organization. Leave empty if not applicable.

    • base_url (default: 'https://api.openai.com/v1'): The base URL for the OpenAI API. You can use any OpenAI-compatible API by changing this value.

    • stop (default: empty list): A list of strings where the API will stop generating further tokens.

    • temperature (default: 0): The sampling temperature for generating responses.

    • max_retries (default: 3): The maximum number of retries for failed requests.

    • seed (default: -1): The random seed for generating responses. Set to -1 for no seed.

    • frequency_penalty (default: 0): The frequency penalty to apply to generated tokens.

    • presence_penalty (default: 0): The presence penalty to apply to generated tokens.

Addon Providers

Anthropic

  • Availability: Anthropic Addon

  • Name: anthropic

  • Options:

    • model (required): The name of the Anthropic language model to use.

    • timeout (default: 15): The request timeout in seconds.

    • max_tokens (default: 0): The maximum number of tokens to generate in the response.

    • temperature (default: 0): The sampling temperature for generating responses.

    • max_retries (default: 3): The maximum number of retries for failed requests.

    • stop (default: empty list): A list of strings where the API will stop generating further tokens.

Azure OpenAI

  • Availability: Azure OpenAI Addon

  • Name: azure-openai

  • Options:

    • deployment_name (required): The name of the Azure OpenAI deployment (model) to use.

    • timeout (default: 15): The request timeout in seconds.

    • max_tokens (default: 0): The maximum number of tokens to generate in the response.

    • max_retries (default: 3): The maximum number of retries for failed requests.

    • stop (default: empty list): A list of strings where the API will stop generating further tokens.

Google AI Studio

  • Name: google-aistudio

  • Options:

    • model (required): The name of the Google AI Studio language model to use.

    • timeout (default: 15): The request timeout in seconds.

    • max_tokens (default: 0): The maximum number of tokens to generate in the response.

    • temperature (default: 0): The sampling temperature for generating responses.

    • max_retries (default: 3): The maximum number of retries for failed requests.

HuggingFace

  • Availability: HuggingFace Addon

  • Name: huggingface

  • Options:

    • model (required): The name of the HuggingFace language model to use.

    • timeout (default: 15): The request timeout in seconds.

    • max_tokens (default: 0): The maximum number of tokens to generate in the response.

    • temperature (default: 0): The sampling temperature for generating responses.

    • max_retries (default: 3): The maximum number of retries for failed requests.

Mistral AI

  • Availability: Mistral AI Addon

  • Name: mistralai

  • Options:

    • model (required): The name of the Mistral AI language model to use.

    • timeout (default: 15): The request timeout in seconds.

    • max_tokens (default: 0): The maximum number of tokens to generate in the response.

    • temperature (default: 0): The sampling temperature for generating responses.

    • max_retries (default: 3): The maximum number of retries for failed requests.

    • base_url (default: 'https://api.mistral.ai/v1'): The base URL for the Mistral AI API. You can use any Mistral AI-compatible API by changing this value.

Was this helpful?