[Qwen3.8-Flash Officially Launches on B.AI, API Official Group Fully Open for Free Access] According to Deep Tide TechFlow, on August 27, the one-stop AGI infrastructure platform B.AI announced that Alibaba's latest open-source Qwen3.8-Flash model has officially launched on the platform and is fully open for free access via the API official group. As a forward-looking preview of the Qwen4 architecture, Qwen3.8-Flash is a multimodal MoE model with 125B total parameters and 51B N-gram embeddings, activating only 6B parameters per token. The model integrates GDN and QSA hybrid attention mechanisms for the first time, balancing long-text processing efficiency with information retrieval accuracy. It natively supports 262K context and can be extended to 1M through YaRN. Starting today, developers can access this model at zero cost via the B.AI API official group, enabling seamless handling of multimodal tasks such as programming assistance, intelligent workflows, and document analysis. B.AI continues to lower the barriers to AI application with inclusive computing power, allowing more developers to leverage top-tier open-source models. Experience it for free now: https://chat.b.ai/chat