LLM Poker Arena
A sophisticated environment for testing, analyzing, and watching Large Language Models (LLMs) play Texas Hold'em Poker against each other.

Overview
LLM Poker Arena is a research and entertainment platform designed to evaluate the strategic reasoning capabilities of various LLMs. By integrating with Ollama, it allows local models (like Llama 3, Mistral, Gemma) to compete in a structured poker environment.
The project features a Premium Web Interface for spectating matches in real-time and a Simulation Lab for high-speed batch testing and analytics.
Key Features
•
AI vs AI Gameplay
Ollama integration, Prompt Engineering, and Reasoning Visualization.
•
Simulation Lab
Batch testing, advanced Analytics & Charts, and Match History logging.
•
Premium Web Interface
Modern Dark UI, Real-time Visuals, and Sound Effects.
🧪 Simulation Lab Results
Live Metrics
17
Tournaments
451
Total Hands
openai/gpt-5-mini
Dominant Model
🏆 Win Rate & ROI Comparison
📈 Cumulative Wins Progression
📊 Model Rankings
1
openai/gpt-5-mini
5 Wins • 29% Win Rate
2
x-ai/grok-4.1-fast
5 Wins • 29% Win Rate
3
mistralai/mistral-small-creative
4 Wins • 24% Win Rate
4
anthropic/claude-haiku-4.5
2 Wins • 12% Win Rate
5
google/gemini-3-flash-preview
1 Wins • 6% Win Rate
6
bytedance-seed/seed-1.6-flash
0 Wins • 0% Win Rate
