Tencent Hunyuan launches Hy4 flash for lower-cost AI tasks
Tencent Hunyuan released Hy4 flash, a multimodal language model with 224B total parameters, 12B active parameters per token, and a 1M-token context window. According to 36氪, API pricing is RMB 0.5 per million input tokens and RMB 2 per million output tokens, with cached input priced at RMB 0.02; IT之家 reports that the model is already available through Tencent Cloud TokenHub, WorkBuddy, Yuanbao, WeCom, QQ, ima, and Marvis.
Hy4 flash gives developers and product teams a lower-priced multimodal option with a long context window, while its stated cache efficiency and task-cost reductions may broaden its use in everyday AI workflows.

智东西 作者 | 江宇 编辑 | 心缘
来来来,先看一段MV。
穿着黑色皮衣的黄仁勋,变成了像素小人,在蔡依林的《日不落》中登台亮相。唱到“爱的明信片”时,黄仁勋掏出的是一颗绿色GPU芯片。
熟悉的黑皮衣、熟悉的动作,也算还原出几分英伟达GTC的味道。
这支MV当然谈不上精致。黄仁勋的动作有些僵硬,画面细节也比较粗糙,连歌词字幕都没能准确对上背景音乐。
#Tencent Hunyuan#Hy4 flash#Hy4#Hy3#WorkBuddy#Tencent Cloud TokenHub#Yuanbao#WeCom#QQ#ima#Marvis#OpenRouter