[Qwen Releases Max-Level Flagship Weights for the First Time, Qwen3.8 Officially Available for Download] According to monitoring by Beating, Alibaba's Qwen has officially released the Max-level model weights for Qwen3.8, now available on Hugging Face and ModelScope. The model is named Qwen3.8-2.4T-A95B, featuring 2.4T parameters with 95B activated per inference. This marks the first time Qwen has opened its Max-level flagship weights. However, the open version is not entirely identical to the cloud-based Qwen3.8-Max. The official model card specifies that the open version natively supports 262K context length, expandable to approximately 1M, and allows for adjustable inference intensity. The current version only supports text input and requires the Thinking mode to be enabled. The cloud-based Qwen3.8-Max additionally supports visual input, non-Thinking mode, a default 1M context length, and built-in official tools. The license has also changed from Apache 2.0 to Qwen's custom license. Regular users can still download, modify, and deploy the model, but certain large-scale commercial uses will require separate authorization. Reuters previously reported that Alibaba plans to impose additional fees or licensing conditions for large-scale commercial users of Qwen3.8-Max. The previously announced Qwen3.8-27B has not yet been released. [Original Link]