Qwen3.8-27B 开源，支持本地部署及商用
Qwen3.8-27B 开源，性能超越 Qwen3.7-Plus，支持家用显卡本地运行，是值得关注的模型发布。
Qwen3.8-27B 正式开源，采用 Apache 2.0 许可，可免费下载、部署和商用。它是一款原生多模态 Dense 模型，原生上下文长度 262K，可通过 YaRN 扩展到 1M 令牌。相比前代，编程和办公性能显著提升，并新增推理深度调节功能。
正文摘录
Breaking: Qwen3.8-27B Is Now Open Source — It Runs on Consumer GPUs Today, we officially open-source the Qwen3.8 series models. All developers, research institutions, and enterprises can freely download, deploy, and use them. https://huggingface.co/collections/Qwen/qwen38 https://www.modelscope.cn/collections/Qwen/Qwen38 The 27B (27 billion parameter) scale is the most in-demand model size in the global AI community — and Qwen3.8-27B just went live as open source. It's easy to deploy, fast, and practical. Qwen3.8-27B is a natively multimodal dense model with 262K native context, extensible to 1M tokens via YaRN. Compared to Qwen3.6-27B, the new model delivers significant gains in programming and office productivity scenarios, even surpassing Qwen3.7-Plus. It also adds a reasoningeffort fea…