Overview

Self-Hosting Documentation Access Granted

LlamaCloud supports multiple LLM models through different provider access methods to power its document parsing, extraction, and AI capabilities. This section provides guidance on configuring and choosing between different model providers for your self-hosted deployment.

Supported Models and Providers

Model Family	Developer Direct (Simple Setup)	Enterprise Cloud (Advanced Features)
OpenAI GPT GPT-4o, GPT-4.1, GPT-5, GPT-5	OpenAI API	Azure OpenAI
Anthropic Claude Claude 4.0 Sonnet, Claude 3.5 Haiku, Claude 3 Opus	Anthropic API	AWS Bedrock
Google Gemini Gemini 2.5 Pro, Gemini 2.5 Flash, Gemini 2.0 Flash	Google Gemini API	Google Vertex AI

Configuration Methods

External Secrets (Recommended)

Configure LLM credentials using Kubernetes secrets and reference them in your Helm values:

config:
  llms:
    openai:
      secret: <your-openai-secret>
    anthropic:
      secret: <your-anthropic-secret>
    gemini:
      secret: <your-gemini-secret>
    azureOpenAi:
      secret: <your-azureOpenAi-secret>
    awsBedrock:
      secret: <your-bedrock-secret>
    googleVertexAi:
      secret: <your-vertex-secret>

Helm Values Configuration (Legacy)

Some providers support direct configuration in Helm values (being deprecated):

backend:
  config:
    openAiApiKey: "your-api-key"

Next Steps

Choose your LLM provider and follow the detailed setup instructions:

Troubleshooting

Verification Steps

After configuration, verify your setup by:

Using the LlamaCloud admin UI to confirm available models
Testing with a simple parsing or extraction task

Common Issues

Model not available: Check provider documentation for model availability in your region
Authentication failures: Verify API keys and permissions
Rate limiting: Monitor usage and implement appropriate quotas