Azure AI API Script

A simple bash script for making requests to Azure AI APIs with token usage and cost tracking.

Features

• Make single or batch requests to Azure AI endpoints
• Process multiple requests from an input file
• Track token usage and calculate costs
• Display performance metrics (tokens per second, response time)
• Formatted output for easy reading

Requirements

• Bash shell environment
• curl for making HTTP requests
• jq for parsing JSON responses
• bc for mathematical calculations

Usage

Single Request

./invoke.sh --endpoint_url <url> --api_key <key> [--content <message>] [--input_cost <cost>] [--output_cost <cost>]

Batch Processing

./invoke.sh --endpoint_url <url> --api_key <key> --input_file <path> [--input_cost <cost>] [--output_cost <cost>]

Parameters

Parameter	Description	Default
--endpoint_url	Azure AI API endpoint URL	(Required)
--api_key	Your API key	(Required)
--content	Message content for single request	"Tell me a story"
--input_file	Path to file with one request per line	None
--input_cost	Cost per 1,000 input tokens	0.00013
--output_cost	Cost per 1,000 output tokens	0.00052

Example

# Single request
./invoke.sh --endpoint_url "https://your-endpoint.models.ai.azure.com/chat/completions" --api_key "$API_KEY" --content "Write a haiku about programming" --input_cost 0.00015 --output_cost 0.0006

# Batch processing
./invoke.sh --endpoint_url "https://your-endpoint.models.ai.azure.com/chat/completions" --api_key "$API_KEY" --input_file "requests.txt" --input_cost 0.00015 --output_cost 0.0006

Output

The script provides formatted output including:

• Request content
• Full API response
• Token usage statistics
• Cost calculation
• Response time and tokens per second

Notes

• Modify the input/output costs to match your specific model pricing
• For batch processing, each line in the input file is treated as a separate request