AI 导读
Flam Falcon，Visual Agents 背后的 26B 混合专家模型，可在 30ms 内返回首个 token，专为印度语言优化；而 Flam Fable 则能以原生 alpha 视频形式，从文本或图像提示为 Airboards 生成 3D 资产。 来看看 👀 https://t.co/p2S7nxXt23
正文
Flam Falcon, the 26B mixture-of-experts model behind Visual Agents, can return the first token in 30ms and is specialized for Indic languages, while Flam Fable generates 3D assets for Airboards as native alpha video from a text or image prompt.
Take a look 👀
https://t.co/p2S7nxXt23
来源：@testingcatalog · x.com