Alibaba’s rapid Qwen3.8 rollout in August–September 2026, headlined by the August 3 launch of the 2.4T-parameter Qwen3.8-Max MoE model (95B active parameters, 1M context) via Alibaba Cloud APIs, has shaped trader views on the next dedicated Plus-tier release at 3.8 or higher. The series added open-weight variants including a 27B Apache-licensed multimodal model and efficiency-focused releases such as Qwen3.8-Flash-Next, while maintaining strong benchmark gains in coding and agentic tasks amid competition from DeepSeek, Moonshot, and GLM models. At the late-September Apsara Conference, Alibaba confirmed Qwen4 is in training with 5–10T parameter targets for later variants, and recent QwenGyre agent-training advances underscore ongoing internal iteration. Traders monitor official channels for any Plus-branded 3.8+ announcement or Qwen4 preview as the key near-term catalyst.