Alibaba’s August 2026 launch of Qwen3.8-Max—a 2.4-trillion-parameter mixture-of-experts model with 95 billion active parameters, 1-million-token context, and multimodal capabilities—set a high bar, followed by open weights and incremental checkpoints like the September Qwen3.8-Max-0902 snapshot. No official 3.9+ flagship has been announced, and September’s Apsara conference confirmed Qwen4 training is underway, with larger 4.5/5 variants projected at 5–10 trillion parameters. This shift, alongside competition from Moonshot’s Kimi K3 and DeepSeek models plus post-training tools like October’s QwenGyre framework, has kept near-term release odds low on Polymarket. Traders are watching for benchmark updates, Alibaba earnings, or developer conferences that could accelerate a 3.9+ drop or clarify timelines.