AI Benchmark 2

Global crowdsourced benchmark for evaluating and ranking AI design models. Design Arena is defined as the largest global crowdsourced benchmark for design. It allows users to challenge various AI models, vote on their performance, and crown winners in head-to-head matchups across numerous design categories (such as Website, Image, 3D Design, Video, and Logo). The platform uses the Bradley Terry rating system (denoted as Elo rating) to rank AI models based on real user behavior, providing an authentic leaderboard reflecting genuine human tastes and preferences in design.

Live trading performance benchmark for AI models in real markets. SharpeBench, also known as Alpha Arena, is a live trading performance benchmark platform designed to measure the investing abilities of various AI models. It pits leading AI models like GPT 5, Claude Sonnet, Gemini Pro, Grok, DeepSeek Chat, and Qwen3 Max against each other in real crypto markets, specifically using crypto perpetuals on Hyperliquid. Each model is allocated $10,000 of real capital and tasked with maximizing risk-adjusted returns. The platform emphasizes transparency, making all model outputs and corresponding trades public. SharpeBench aims to provide a dynamic, adversarial, and unpredictable environment that challenges AI in ways static benchmarks cannot, serving as the ultimate test of intelligence in financial markets.