Poker Arena

A comprehensive AI poker tournament system where multiple AI models compete in Texas Hold'em poker. Run single games with a live web interface or execute large-scale parallel tournaments with detailed statistical analysis.

Project Structure

poker-arena/
├── cmd/
│   └── poker-arena/
│       └── main.go            # Application entry point with CLI support
├── internal/
│   ├── ai/
│   │   └── client.go          # AI model communication
│   ├── game/
│   │   ├── game.go            # Core game logic with ID support
│   │   └── actions.go         # Player actions (bet, fold, etc.)
│   ├── poker/
│   │   ├── deck.go            # Card deck management
│   │   └── hand.go            # Hand evaluation with safety checks
│   ├── server/
│   │   └── server.go          # HTTP server and API endpoints
│   └── tournament/
│       ├── manager.go         # Parallel game coordination
│       └── exporter.go        # CSV export functionality
├── pkg/
│   └── models/
│       ├── game.go            # Game data structures
│       ├── tournament.go      # Tournament aggregation models
│       └── config.go          # CLI configuration
├── index.html                 # Frontend web interface
├── go.mod                     # Go module dependencies
└── go.sum                     # Go dependency checksums

Quick Start

Prerequisites

• Go 1.24.2 or higher
• AI API keys from supported providers (OpenAI, Google, Anthropic, or OpenRouter for testing multiple models)

Installation

1. Clone the repository:

   git clone https://github.com/MikeLuu99/poker-arena.git
   cd poker-arena

2. Install dependencies:

   go mod download

3. Create a .env file with your AI API keys:

   # Add your AI API keys here
   OPENAI_API_KEY=your_openai_key
   GOOGLE_API_KEY=your_google_key
   ANTHROPIC_API_KEY=your_anthropic_key
   
   # Alternative: Use OpenRouter for testing different models
   # OPENAI_API_KEY=your_openrouter_key  # Use OpenRouter key for all OpenAI models
   # GOOGLE_API_KEY=your_openrouter_key  # Use OpenRouter key for Google models
   # ANTHROPIC_API_KEY=your_openrouter_key  # Use OpenRouter key for Anthropic models
   
   PORT=3000

4. Run the application:

   # Single game with web interface
   go run cmd/poker-arena/main.go
   
   # 3 parallel games with web interfaces on ports 3000-3002
   go run cmd/poker-arena/main.go --games 3 --with-servers --verbose
   
   # 10 parallel games batch mode with CSV export
   go run cmd/poker-arena/main.go --games 10 --output results.csv --no-server

Command Line Options

Flag	Short	Description	Default
--games	-g	Number of parallel games to run	1
--output	-o	CSV output file path	poker_results.csv
--no-server		Disable web server (batch mode)	false
--with-servers		Enable web servers for parallel games	false
--verbose	-v	Enable detailed logging	false
--port		Base web server port for parallel games	3000
--help	-h	Show help information

Environment Configuration

Variable	Description	Required
OPENAI_API_KEY	OpenAI API key for GPT models (or OpenRouter key)	Yes
GOOGLE_API_KEY	Google API key for Gemini (or OpenRouter key)	Yes
ANTHROPIC_API_KEY	Anthropic API key for Claude (or OpenRouter key)	Yes
PORT	Web server port (overridden by --port flag)	No

AI Models

• Google Gemini 2.5 Flash: Advanced reasoning and strategic play
• OpenAI GPT-5 Nano: Efficient decision-making model
• OpenAI GPT OSS 120B: Large-scale language model
• Anthropic Claude 3.5 Haiku: Fast and strategic AI player

Note: You can use OpenRouter API to access different models for testing and comparison. Simply use your OpenRouter API key and specify model names in the format expected by OpenRouter.