Aravind Srinivas · @AravSrinivas · X·2026-08-26 04:08·18小时前
AI 导读

Perplexity 发布本地优先智能体 Portable Computer,搭载端侧 27B 模型,在真实知识工作基准上得分 82.6%,超越开源方案 Pi 和 Hermes;后训练版 PPLX 27B 达 85.4%。Perplexity CEO 感谢英伟达在 DGX Spark 上的合作,并推动开放生态以支持高性价比开源权重模型、推理框架及统一内存硬件。

Aravind Srinivas@AravSrinivas
33AI 编辑部评分,满分 100
2026-08-26 04:08· 18小时前
AI 导读

Perplexity 发布本地优先智能体 Portable Computer,搭载端侧 27B 模型,在真实知识工作基准上得分 82.6%,超越开源方案 Pi 和 Hermes;后训练版 PPLX 27B 达 85.4%。Perplexity CEO 感谢英伟达在 DGX Spark 上的合作,并推动开放生态以支持高性价比开源权重模型、推理框架及统一内存硬件。

Big thanks to @nvidia for working together with us and supporting this research on DGX Spark as well as enabling an open ecosystem around cost-effective open-weight models, inference frameworks, and hardware with unified memory.

PerplexityNew research: Portable Computer is a local-first agent for private and cost-effective work. With an on-device 27B model, our harness scores 82.6% on real knowle...

来源:Aravind Srinivas· x.com