Skip to main content

Step 4: LiteLLM Proxy Model Configuration

The core of the LiteLLM Proxy configuration is the model list, which defines the LLM models the proxy will manage. Sample configuration files are provided in the Helm chart for each major cloud provider.

Configuration File Locations

  • AWS Bedrock: litellm/config/litellm-aws-config.yaml
  • Azure OpenAI: litellm/config/litellm-azure-config.yaml
  • Google Vertex AI: litellm/config/litellm-gcp-config.yaml

Regional Configuration Overview

The sample configurations specify default regions for accessing the models. The table below provides a summary.

ProviderModelDefault Region
AWS Bedrockclaude-3-5-sonnetus-west-2
AWS Bedrockclaude-3-5-sonnet-v2us-west-2
AWS Bedrockclaude-3-7us-west-2
AWS Bedrockclaude-4-sonnet-1mus-west-2
AWS Bedrockclaude-4-sonnetus-west-2
AWS Bedrockclaude-4-sonneteu-central-1
AWS Bedrockclaude-4-opusus-west-2
AWS Bedrockclaude-4-1-opusus-west-2
AWS Bedrocktitanus-west-2
Google Vertex AIgemini-2.0-flashus-central1
Google Vertex AIgemini-2.5-prous-central1
Google Vertex AIclaude-sonnet-v2-vertexus-east5
Google Vertex AIclaude-sonnet-3-7-vertexus-east5
Google Vertex AItext-embedding-005us-central1
Azure OpenAIgpt-4o-2024-08-06eastus2
Azure OpenAIgpt-4o-2024-11-20eastus2
Azure OpenAIgpt-4o-minieastus2
Azure OpenAIgpt-4.1eastus2
Azure OpenAIgpt-4.1-minieastus2
Azure OpenAIgpt-5-2025-08-07eastus2
Azure OpenAIgpt-5-mini-2025-08-07eastus2
Azure OpenAIgpt-5-nano-2025-08-07eastus2
Azure OpenAIgpt-4-visioneastus2
Azure OpenAIo3-minieastus2
Azure OpenAIo1eastus2
Azure OpenAIo3-2025-04-16eastus2
Azure OpenAIo4-mini-2025-04-16eastus2
Azure OpenAIcodemie-text-embedding-ada-002eastus2

Customizing Regions

You can customize the regions where your models are accessed by modifying the corresponding configuration files.

AWS Bedrock

To use a different AWS region, modify the aws_region_name parameter in the model's configuration.

model_list:
- model_name: claude-4-sonnet
litellm_params:
aws_region_name: us-west-2 # Change this to your preferred region, e.g., "eu-central-1"

Azure OpenAI

For Azure, the region is determined by the endpoint URL in the api_base parameter.

model_list:
- model_name: gpt-4o-2024-08-06
litellm_params:
model: azure/gpt-4o-2024-08-06
api_base: https://your-resource.openai.azure.com/ # Your Azure OpenAI endpoint defines the region
litellm_credential_name: default_azure_openai_credential

Google Vertex AI

For Google Vertex AI, the method for region configuration depends on the model family:

Gemini Models

Region is set globally in the litellm_settings section.

litellm_settings:
vertex_project: os.environ/VERTEX_PROJECT
vertex_location: 'us-central1' # Change this to your preferred region for Gemini models

Claude Models on Vertex AI

Region and project are specified directly in each model's configuration.

model_list:
- model_name: claude-sonnet-3-7-vertex
litellm_params:
model: vertex_ai/claude-3-7-sonnet
vertex_ai_project: os.environ/VERTEX_PROJECT
vertex_ai_location: 'us-east5' # Change this to your preferred region for this specific model

Next Steps

Continue to Deployment.