nix/hosts/open-webui/docker.nix

{ config, pkgs, ... }:
let
  litellmSettings = {

    general_settings = {
      proxy_batch_write_at = 60; # Batch write spend updates every 60s
      database_connection_pool_limit = 10; # Limit the number of database connections
      disable_spend_logs = true; # Turn off writing each transaction to the DB
      disable_error_logs = true; # Turn off writing LLM exceptions to DB
      allow_requests_on_db_unavailable = true; # Allow requests if DB is unavailable
    };

    environment_variables = {
      LITELLM_MODE = "Production";
    };

    credential_list = [
      {
        credential_name = "dp_azure_openai_credential";
        credential_values = {
          api_base = "os.environ/AZURE_API_BASE_OPENAI";
          api_key = "os.environ/AZURE_API_KEY_OPENAI";
        };
        credential_info = {
          description = "Azure OpenAI credentials for DP";
        };
      }
      {
        credential_name = "dp_azure_ai_credential";
        credential_values = {
          api_base = "os.environ/AZURE_API_BASE_AI";
          api_key = "os.environ/AZURE_API_KEY_AI";
        };
        credential_info = {
          description = "Azure AI credentials for DP";
        };
      }
    ];

    model_list = [
      {
        model_name = "text-embedding-3-large";
        litellm_params = {
          model = "azure/text-embedding-3-large";
          litellm_credential_name = "dp_azure_openai_credential";
        };
        model_info = {
          mode = "embedding";
        };
      }
      {
        model_name = "text-embedding-3-small";
        litellm_params = {
          model = "azure/text-embedding-3-small";
          litellm_credential_name = "dp_azure_openai_credential";
        };
        model_info = {
          mode = "embedding";
        };
      }
      {
        model_name = "GPT-3.5 Turbo";
        litellm_params = {
          model = "azure/gpt-35-turbo";
          litellm_credential_name = "dp_azure_openai_credential";
        };
      }
      {
        model_name = "GPT-4o";
        litellm_params = {
          model = "azure/gpt-4o";
          litellm_credential_name = "dp_azure_openai_credential";
        };
      }
      {
        model_name = "GPT o3 Mini";
        litellm_params = {
          model = "azure/o3-mini";
          litellm_credential_name = "dp_azure_openai_credential";
        };
      }
      {
        model_name = "GPT-4o Mini";
        litellm_params = {
          model = "azure/gpt-4o-mini";
          litellm_credential_name = "dp_azure_openai_credential";
        };
      }
      {
        model_name = "Dall-e 3";
        litellm_params = {
          model = "azure/dall-e-3";
          litellm_credential_name = "dp_azure_openai_credential";
        };
        model_info = {
          mode = "image_generation";
        };
      }
      {
        model_name = "azure-openai-4o-audio";
        litellm_params = {
          litellm_credential_name = "dp_azure_openai_credential";
          model = "azure/gpt-4o-audio-preview";
        };
      }
      {
        model_name = "DeepSeek-R1";
        litellm_params = {
          litellm_credential_name = "dp_azure_ai_credential";
          model = "azure_ai/deepseek-r1";
        };
      }
    ];
  };
in
{

  age.secrets = {
    open-webui.file = ../../secrets/open-webui.age;
  };

  my = {

    virtualisation.docker.enable = true;

    utils = {
      commons.enable = true;
      commons.gc.enable = true;
      lxc-standard.enable = true;
    };

    virtualisation.proxmox.enable = true;
  };

  system.stateVersion = "24.11";
}