The Qwen3.5 native vision-language Flash models are built on a hybrid architecture that integrates a linear attention mechanism with a sparse mixture-of-experts model, achieving higher inference efficiency. Compared to the...
模态
文本图像视频
输入 / 输出价格
₩114 / ₩456
1M
上下文
100万
发布日期
2026年2月26日
暂无运行时间数据。