In a real-money AI crypto trading showdown, Elon Musk’s Grok, DeepSeek, and Anthropic’s Claude Sonnet 4.5 are top performers, each generating over 25% returns. OpenAI’s GPT-5 and Google’s Gemini 2.5 Pro suffered staggering losses of more than 28% in the same period in the “Alpha Arena” competition.

Given $10,000 to trade on Hyperliquid exchange, AI models aim to maximize risk-adjusted returns autonomously, trading Bitcoin, Dogecoin, and Solana. Season 1 from October 17 to November 3 sees Grok and DeepSeek leading, with GPT-5 down about 29% due to a cautious strategy.

GPT-5’s conservative approach led to minimal gains but protected from significant losses. Grok’s success suggests general-purpose AI can navigate markets independently. DeepSeek’s specialized financial data hints at a future with new AI-driven financial analysis.

AI trading enthusiasts see LLMs analyzing vast datasets as the future of trading, unlocking new alpha. However, Gemini’s losses highlight AI trading risks, including lack of transparency and reliability. Concerns about market instability and flash crashes exist with multiple AI agents reacting to events in a correlated way.

Wall Street cautiously explores AI adoption, mainly for low-risk tasks with human assistance. Despite potential benefits, AI trading’s unpredictability and risks, as demonstrated by Gemini’s erratic performance, keep financial institutions wary of widespread adoption.

Read more at Yahoo Finance: DeepSeek and Grok Are Cashing In as Gemini Implodes