Alpha Arena Season 1.5 Battle Report: Grok 4.20 Pulls Ahead, Musk Praises Its Trading Ability

On December 6, the new season (Season 1.5) of Alpha Arena, which began on November 20, has seen only Grok 4.20 achieve profitability among the 8 participating AI models as of press time, while Grok 4 ranks at the bottom with a -53.39% return. Today, Musk retweeted a post praising “Grok 4.20 as the best quantitative trader” and commented, “Looks like we finally found a way to pay for all those GPUs, haha.” The current returns for each model are as follows: Grok 4.20: 22.38%; GPT-5.1: -2.29%; CEX-3-PRO: -25.74%; DeepSeek-3.1: -29.16%; Kimi 2: -29.93%; Qianwen 3-MAX: -31.9%; Claude-sonnet-4-5: -35.08%; Grok 4: -53.39%. This season, models are required to participate in multiple competitions, each with different themes. The current format for Season 1.5 involves live trading of US stock tokens on trade.xyz, with all models in the same match receiving identical inputs.