
HEADROOM — how fast is your machine, really?
测量GPU为本地AI的真实内存带宽上限(30秒)。
Ar5en1c · HN
Headroom – measure your GPU's true bandwidth ceiling for local AI
完整作品展
技术栈
60 projects

测量GPU为本地AI的真实内存带宽上限(30秒)。
Ar5en1c · HN
Headroom – measure your GPU's true bandwidth ceiling for local AI

为语音AI代理提供实时记忆和上下文处理,延迟低至10ms。
@SouravDaaa · X
- End to end personalised memory retrieval in 10ms. Would love some feedback here


Recall 是用于工作会话的 AI 记忆。它在您工作时静静地运行,将您的日程整理成项目、线程和任务,这样您可以一键返回任何工作片段
@vitverb · X
hi :) One tap back to the work you meant to finish. No tasks to manage. Nothing to set up. Local-first

为 Claude 等 LLM 添加永久记忆,延续对话不重复解释。
@Nikborneklint · X
Shipped today: AEGIS Code's Terminal picker now includes @xai's grok CLI alongside AEGIS Code & Claude Code — all sharing one memory bridge via MCP. Local memory's free forever. #grokbuild @elonmusk

Deep Nostalgia AI 将老照片转为生动视频,保留珍贵回忆。
Deep Nostalgia AI — 借助人工智能照片动画,将家庭照片转变为引人入胜的动画视频,为珍贵的回忆注入活力的网站


一次存储共享记忆,在AI系统和团队中复用。
Repeater22746 · HN
ContextVault – Shared memory layer for your AI and your team

在Second Mind中保存书签、视频和笔记,AI知识库会总结内容并关联想法。
@vhm205 · X
Too many news articles, too many bookmarks, too many notes, but I never open them again because I don't know where to find them

与能记忆并学习的 AI Veiled Prime 进行持续对话。
@Vematrex · X
5M views. 13K shares. Building Veiled Prime at The most insightful AI we know how to make. Built for people doing real work. One conversation. Real memory. No resets.


保存并浏览你想尝试的餐厅,帮助你决定去哪里用餐。
@danielendara · X
Sure! let's connect! My latest creation is