Recent benchmark leadership by Xiaomi’s MiMo-V2.6-Pro, Alibaba’s Qwen3.8 Max, and Moonshot’s Kimi K3, alongside DeepSeek’s V4.1 Flash, has kept the second-tier race fluid as Chinese labs close the gap to U.S. frontier models to roughly 3 percent on key indices. Traders assign Alibaba the highest implied probability among named contenders because of its broad open-weight ecosystem, enterprise agent launches like Qianwen Go, and consistent top-three placement across Artificial Analysis and LiveBench. Xiaomi’s rapid model cadence and low-cost MIT-licensed releases drive its second-place odds, while Moonshot’s large funding round and long-context strengths support its share. DeepSeek’s $12 billion-plus raise and record token usage create upward pressure on its standing, though lower listed probabilities reflect a crowded field where new releases or adoption shifts before month-end could reorder the consensus.