LLM Holdem - Play poker with LLMs, or watch them play against each other

What it does

LLM Holdem is a web-based platform where users can play Texas Hold'em poker against large language models (LLMs) or spectate as multiple LLMs compete in automated games. The interface displays a poker table with player positions, chip counts, and betting actions. Supported models include Claude (Opus 4.5, Sonnet 4.5), Grok (4 Fast), GPT (5.2), DeepSeek (V3.2), and Gemini (3 Flash). Users can create or join rooms, and the system logs all actions for review.

Who it is for

This platform is designed for AI enthusiasts, poker players, and researchers interested in observing or testing the decision-making capabilities of different LLMs in a competitive, strategic game. It also appeals to anyone curious about how AI models handle bluffing, risk assessment, and game theory.

Why it matters

Poker is a classic test of strategic reasoning, bluffing, and probabilistic thinking. By pitting various LLMs against each other, LLM Holdem provides a transparent, interactive benchmark for comparing model performance in a dynamic environment. It also offers entertainment and educational value, allowing users to see firsthand how different AI personalities approach the same hand.

Launch signal

LLM Holdem was launched on Hacker News as a "Show HN" project. The website is live and functional, with a leaderboard and room creation features. The project appears to be in an early stage, with limited documentation or community features.

Brand and naming

The name "LLM Holdem" directly combines "LLM" (large language model) with "Hold'em" (Texas Hold'em poker), clearly communicating the product's purpose. The branding is straightforward and functional, with a simple logo and interface. The domain llmholdem.com is memorable and easy to type.