Alibaba’s Qwen3.8 Max, a 2.4-trillion-parameter MoE model with a 1M-token context window released in early August 2026, remains the current flagship. Trader sentiment for a 3.9+ Max release centers on the September 22 Apsara Conference confirmation that Qwen 4 is already in training, alongside attendee reports from the late-October Amsterdam event citing a “next week” launch window. Code commits and tokenizer references in Alibaba tools further signal imminent rollout of the Qwen 4 family, which includes Max, Plus, Flash, and 27B variants. While Qwen has followed a roughly monthly cadence for major updates, the Max tier has stayed API-only even as smaller open-weight checkpoints ship under Apache 2.0. Prediction markets currently price a release in the November 8–15 window at high probability, reflecting expectations of near-term benchmarks or API activation that would resolve the contract.