OCTOBER 24TH MESSAGE, YESTERDAY, AT AI Large ModelThe real dealQwen3-Max, under Ali Baba's banner, won the first-ever champion with an efficient strategy and precision, breaking DeepSeek's long lead。
This competition, initiated by the American Artificial Intelligence Research Laboratory, Nof1.ai, provides real money of $10,000 each to the six top American models for real-time investments on the Hyperliquid Exchange. The participation models include Claude 4.5 Sonnet, DeepSeek V3.1 Chat, Gemini 2.5 Pro, GPT-5, Grok 4, and Qwen3-Max。
All models are traded under the same warning and data entry conditions, with the objective of " maximizing risk adjusted benefits". The model needs to be independently produced for Alpha generation, warehouse decision-making, trading timing and risk management, and relevant transaction records are publicly available on the nof1.ai network。
By 10:30 a.m. Beijing time on October 23rd, Qwen ' s total account was valued at $11,252.34, surpassing DeepSeek ' s $10,868.84, making it the best-performing AI transaction model. Qwen's manoeuvre strategy emphasizes the "Shap ratio" indicator, and quick capture of trading opportunities is key to its reversal。
It is worth noting that only two models, all of which are national AI, currently achieve positive returns. The remaining four overseas models are in deficit, of which GPT-5 and Gemini 2.5 Pro performed the worst, with current warehouse market value of US$ 3,119 and US$ 4,444, respectively, with cumulative losses of over US$ 6,800 and US$ 5,500, with significant war performance。