Qualcomm launches GenieX to run LLMs on their Windows Laptops

推荐理由

Llama 相关模型动态已经出现，适合跟踪能力变化、生态影响和后续可用性。

AI 摘要

高通推出 GenieX，旨在其 Windows 笔记本电脑上运行大型语言模型。高通在 SDK 方面正在追赶其他主要芯片制造商。GenieX 能够以每秒 20 个 token 的速度在 GPU 或 NPU 上运行 Gemma 4 26B A4B 模型，首个 token 响应时间为 0.5 秒。…

Qualcomm was behind every major chipmaker so they are playing catchup when it comes to SDKs.

https://aihub.qualcomm.com/geniex

I was able to get 20 tok/s running Gemma 4 26B A4B 0.5s for first token running on the GPU or NPU

10 tok/s on the GPU for Qwen 3.6 27B MTP

To use llama.cpp, just get any Q4_0 GGUF model and it will run on the CPU,GPU,NPU

主题标签LlamaQwen模型发布

原始关键词#launches#qualcomm#laptops#windows#geniex#llms

单一来源，暂无交叉验证