A particularly interesting event, I recommend everyone to join in the fun. On October 18, the AI research team nof1, which focuses on finance, conducted an experiment called Alpha Arena, putting six mainstream large models on the same track, allowing them to each manage a sum of $10,000, autonomously placing orders in Hyperliquid's contract market. Real-time profit and loss calculations are made, and the leaderboard is publicly visible. The official site positions it as a benchmark for real-time trading capabilities. Trading situation dashboard: The participating models include DeepSeek, Grok, ChatGPT, Gemini, Claude, Qwen, etc., with trading targets being mainstream coins like BTC, ETH, SOL, BNB, DOGE, and XRP, all starting with the same parameters to minimize human intervention and hint differences. Surprisingly, the one currently making the most profit is actually DeepSeek. It has made just under $3,000. At the bottom is Gemini, which has lost over $4,000. Even CZ commented, stating that AI trading will be a trend, but to form a sustainable advantage, the strategy should ideally be "unique and superior" rather than everyone using the same approach. It seems that in the future, we won't even have to lose money ourselves; AI can take care of that for us. I wonder when we can really live off AI profits? Speaking of this, I suddenly thought that @SentientAGI is also an aggregator of many AI agents, and it currently has cryptocurrency agents, but it can only query cryptocurrency information in real-time. However, I believe that once the full version of Sentient is launched, there will definitely be developers uploading impressive trading AI, because the more people use the developers' open-source agents, the higher the developers' earnings will be. The days of AI online PVP shouldn't be far off. Also, I have a question: why is DeepSeek so strong in trading? Can someone knowledgeable explain?