v0.7.0 — Add Openrouter

[0.7.0] add openrouter

Added

• OpenRouter API Integration: Complete OpenRouter client implementation for live model benchmarking
• Direct API calls to OpenRouter models with --model flag
• Automatic API key management from file or environment variable
• Built-in retry logic with exponential backoff for rate limits and server errors
• Token usage tracking and timing metrics for all API calls
• Batch Benchmarking: Run benchmarks on multiple models simultaneously
• --run-all flag to test all models from models.txt
• --sequential flag for one-at-a-time execution with progress bars
• Parallel execution with ThreadPoolExecutor for faster batch testing
• Automatic skipping of already-tested models to avoid duplicate work
• Real-time progress tracking with tqdm integration
• Enhanced Leaderboard System: Improved tracking and display of benchmark results
• Failed runs tracking with separate display section (Score: -100)
• Medal emojis (🥇🥈🥉) for top 3 performers
• Separate sections for successful vs failed runs
• Multiple runs support per model with timestamp tracking
• Improved Error Handling: Robust error management across all components
• Graceful handling of empty LLM responses with configurable retries
• Rate limit detection and automatic retry with exponential backoff
• Clear error messages for API failures, timeouts, and invalid responses
• Safe file operations with proper exception handling

Changed

• CLI Interface Overhaul: Completely restructured command-line interface
• Default behavior changed to sequential batch mode when no specific action is provided
• --model flag for direct OpenRouter model testing
• --retries parameter for configuring empty response retry attempts
• Enhanced help text with comprehensive usage examples
• Leaderboard Display: Enhanced markdown formatting with better organization
• Separate sections for successful and failed benchmark runs
• Improved table formatting with rank numbers and medal indicators
• Time tracking display for all benchmark runs
• Better visual hierarchy with section headers and spacing
• README Updates: Comprehensive documentation updates
• Added OpenRouter setup instructions for API key configuration
• New usage examples for batch processing and individual model testing
• Updated architecture diagram to include new components
• Enhanced quick start section with practical examples

Improved

• Performance: Optimized batch processing with parallel execution support
• User Experience: Better progress feedback and error messages
• Reliability: Robust retry mechanisms and error recovery
• Scalability: Support for testing multiple models efficiently
• Documentation: Comprehensive examples and setup instructions

Updated

• Leaderboard Results: Latest benchmark results with new model rankings
• Added 8 new model results (4 successful, 4 failed)
• Updated top performers: gemini 3 pro preview (1446.46), deepseek v3.2 (1338.83 points,