Alibaba Cloud has released the open weights for Qwen3.8-Flash-Next, a multimodal model previewing Qwen4. It upgrades attention, residuals, embedding, and optimization to boost capability and reduce costs, with 125B parameters. https://alternativeto.net/news/2026/8/alibaba-cloud-launches-qwen3-8-flash-next-with-new-architecture-and-better-cost-efficiency/