#!/bin/bash set -euo pipefail # Start LiteLLM proxy server for promptfoo example echo "Starting LiteLLM proxy server..." echo "" echo "Required environment variables:" echo " - OPENAI_API_KEY (for GPT models and embeddings)" echo " - ANTHROPIC_API_KEY (for Claude models)" echo " - GOOGLE_AI_API_KEY (for Gemini models)" echo "" # Check if litellm is installed if ! command -v litellm &>/dev/null; then echo "ERROR: LiteLLM is not installed." echo "Install it with: pip install 'litellm[proxy]'" exit 1 fi # Check for at least one API key if [ -z "$OPENAI_API_KEY" ] && [ -z "$ANTHROPIC_API_KEY" ] && [ -z "$GOOGLE_AI_API_KEY" ]; then echo "ERROR: No API keys found. Set at least one of the environment variables above." exit 1 fi echo "Starting proxy on http://localhost:4000..." echo "" # Start the proxy with models used in the example litellm \ --model gpt-4.1 \ --model claude-sonnet-4-6 \ --model gemini-2.5-pro \ --model text-embedding-3-large \ --port 4000