Provider	Template File	Models Supported
OpenAI	`openai.yaml`	GPT-4, GPT-4-Turbo, GPT-3.5-Turbo, o1, o1-mini
Anthropic	`anthropic.yaml`	Claude-3-Opus, Claude-3-Sonnet, Claude-3-Haiku, Claude-3.5-Sonnet
Google	`google.yaml`	Gemini-Pro, Gemini-Ultra, Gemini-Flash
Ollama	`ollama.yaml`	Llama-2, Mistral, CodeLlama, local models
Azure OpenAI	`azure-openai.yaml`	Azure-hosted GPT-4, GPT-35-Turbo
xAI	`xai.yaml`	Grok-Beta, Grok-Vision
Combined Example	`combined-example.yaml`	Multi-provider configuration

Quick Start

Option 1: Include in Main Config

Copy the contents of any template file and paste into your litellm_config.yaml under the model_list: section.

# litellm_config.yaml
model_list:
  # ... existing models ...
  
  # Add OpenAI models
  - model_name: openai/gpt-4-turbo
    litellm_params:
      model: openai/gpt-4-turbo
      api_key: os.environ/OPENAI_API_KEY
    model_info:
      description: "OpenAI GPT-4 Turbo"

Option 2: Use as Separate Config File

Reference external config files using LiteLLM's configuration loading:

# Start LiteLLM with multiple config files
litellm --config litellm_config.yaml --config config/providers/openai.yaml

Option 3: Merge Configurations

Use a YAML merge tool or script to combine configurations:

# Example using yq (install with: pip install yq)
yq eval-all '. as $item ireduce ({}; . *+ $item)' litellm_config.yaml config/providers/openai.yaml > merged_config.yaml

Environment Variables

Each template uses environment variables for API keys. Set these in your .env file:

# OpenAI
OPENAI_API_KEY=sk-...

# Anthropic
ANTHROPIC_API_KEY=sk-ant-...

# Google
GOOGLE_API_KEY=...
GOOGLE_VERTEX_PROJECT_ID=your-project-id

# Azure OpenAI
AZURE_API_KEY=...
AZURE_API_BASE=https://your-resource.openai.azure.com/
AZURE_API_VERSION=2024-02-15-preview

# xAI
XAI_API_KEY=...

Model Naming Convention

Templates follow this naming convention:

provider/model-name - Standard format
provider/deployment-name - For Azure deployments
agent/role - For agent-specific virtual models

Adding a New Provider

Create a new YAML file in this directory
Follow the existing template structure
Include model_info metadata for each model
Document environment variables needed
Add to this README

Template Structure

Each template includes:

model_list:
  - model_name: provider/model-id
    litellm_params:
      model: provider/model-id
      api_key: os.environ/PROVIDER_API_KEY
      # Optional: api_base, api_version, etc.
    model_info:
      description: "Human-readable description"
      max_tokens: 128000
      input_cost_per_token: 0.00001
      output_cost_per_token: 0.00003
      # Optional: mode, supports_vision, supports_function_calling

Usage with OpenClaw

To use these providers with OpenClaw agents:

Add the provider configuration to litellm_config.yaml
Set the required API keys in .env
Restart the LiteLLM container
Optionally update agent model assignments in .env

Example agent assignment:

# Use GPT-4 Turbo for the Coder agent
AGENT_CODER_MODEL=openai/gpt-4-turbo

Troubleshooting

Model Not Found

Ensure the model name in litellm_params.model matches the provider's expected format.

API Key Errors

Verify environment variables are set and accessible to the LiteLLM container.

Rate Limiting

Add rate limiting configuration:

litellm_settings:
  routing_strategy: simple-shuffle
  enable_rate_limiting: true

README.md

LiteLLM Provider Templates

Available Templates