Skip to content

#端侧

0 today

Oct 29Thursday

The Decoder

Google opens Android dev environment to third-party AI agents, launches new Gemini API

At this year's Android Dev Summit, Google introduced a BYOA feature for Android Studio Rabbit 2 and released a new API for connecting on-device agents and apps.

Why it matters: Google is letting third-party coding agents into Android Studio and opening app tool-call interfaces, so agents can use dev tools and app features directly.

Oct 28Wednesday

Hacker News front page

Survival LLM: AI assistant and emergency library that runs on your own computer

Survival LLM 是免费开源的本地 AI 助手与应急资料库,联网完成安装后可离线回答问题,并展示来源文章及对应原文。它支持 Mac 和 Windows,按电脑内存下载适配模型,资料库包含医疗、应急、生存及维修指南和约 930,000 篇 Wikipedia 文章,完整资料库下载量为 9 GB,安装后占用 19 GB。

Hacker News front page

Introducing LFM3: Every Modality for Specialized Intelligence

Liquid AI 发布 LFM3,首批包含 6 款 2B 级模型,以共享骨干覆盖文本、视觉、音频、全模态和决策任务。新骨干采用短卷积、滑动窗口注意力与全局注意力的三混合架构,经过 22T tokens 预训练,文本模型提供 256K tokens 上下文窗口。

Hacker News front page

New 14‑inch MacBook Pro with M6

Apple 发布搭载 M6 的新款 14-inch MacBook Pro,称其端侧 AI 工作负载性能较 M5 最高提升 25 percent。M6 配备 12-core CPU、12-core GPU 和 Dual 16-core Neural Engine,本地 LLM 提示词处理的首 token 性能较 M1 最高提升 13x、较 M5 最高提升 1.3x。

Oct 27Tuesday

Latent Space

Coding From the Himalayas With an Offline AI Agent

Rittika Jindal 在笔记本上通过 OpenCode 和 Ollama 运行 Qwen3.5 9B,断网后完成费用计算程序的修复,原有 5 项测试全部通过。教程给出预下载模型与项目依赖、限制为本地服务、检查断网状态及逐次批准编辑和命令的流程,作者也独立重跑测试并确认测试文件未改动。实测仅涉及一行累加逻辑修复,CSV 报告和日志分析器练习尚未完成离线运行,不能据此判断大型仓库任务的表现。

Hacker News front page

New iMac with M6

苹果推出搭载 M6 芯片的新 iMac,官方称其速度最高达最主流 Intel 版 iMac 的 18 倍、M1 版 iMac 的 12 倍,端侧 AI 性能为 M4 版 iMac 的 5 倍。

TechCrunch · AI

Opera introduces a new web standard to let users deploy AI locally in their browser

Opera 推出 WebAI 网络标准,让用户直接在浏览器本地运行 AI 模型,无需依赖云端。该功能已在 Opera One 提供早期预览,需用户主动启用,Opera GX 和 Opera Neon 将随后支持;兼容模型包括 Google 的 Gemma 和 Mac 上的 Apple 端侧基础模型。WebAI 支持本地文本生成、内容分类和信息关联理解,相关数据留在设备上,模型性能取决于设备硬件。

Hacker News front page

New 14‑inch MacBook Pro with M6

Apple 发布搭载 M6 的 14 英寸 MacBook Pro,强化端侧 AI 能力,美国起售价为 $1,999,10 月 28 日开启预订、11 月 6 日发售。

Hacker News front page

Show HN: Hob - Open-source, on device alternative to Claude Cowork and ChatGPT

Hob 是免费、开源的端侧 AI 工具,可离线使用,支持填写 PDF 和处理文档。它会选择适合用户电脑的前沿开放权重模型并自动完成设置,支持配备至少 8 GB 内存的 Apple silicon Mac、Windows 和 Linux 电脑。下载入口为 hob.chat,项目地址为 https://github.com/FrigadeHQ/hob。

Oct 26Monday

Hacker News front page

SIMD-accelerated computer vision on a $2 microcontroller

开发者利用 ESP32-S3 的 128 位 SIMD 指令实现了加速版 FAST 特征检测器,吞吐量从 5.1MP/s 提升到 11.2MP/s,约提升 220%,处理一帧 QVGA(320x240)图像约需 6ms。

Hacker News front page

New iPad mini

苹果发布新款 iPad mini,搭载 A20 Pro 芯片,配备 8.5 英寸 OLED 屏,峰值亮度较上代提升 2 倍以上,并新增 IP67 防尘抗水。新品 10 月 28 日起在 36 个国家和地区开启预购,11 月 6 日到货,Wi-Fi 版起售价 699 美元,提供 128GB、256GB、512GB 三种容量。

Oct 25Sunday

Hacker News front page

Neural chip decodes interactions between brain regions

EPFL 团队开发出低功耗可植入神经芯片 LogSync,可记录脑活动并提取脑波强度与跨脑区信号同步性等生物标志物,再由机器学习算法区分认知与情绪状态并触发刺激。芯片尺寸约 4x2 mm,功耗低于 200 微瓦,采用对数计算简化脑网络分析。研究基于人类脑记录与大鼠实验验证,论文发表于 IEEE Journal of Solid-State Circuits。

TechCrunch · AI

Block’s new voice model sounds eerily human

Block 发布可本地运行的全双工语音模型 Dessa,支持同时听说,并展示了笑声、停顿和哭声等接近真人的表达。Block 称该模型完全使用人类语音和单张 NVIDIA GPU 训练,知识来自对话数据,情绪表达内置于模型。Dessa 可在 MacBook M4 Max 或更高配置的消费级电脑上运行,目前尚未公开开放,Block 正邀请安全研究人员测试。

Oct 24Saturday

The Decoder

Block's new AI model Dessa was trained entirely on human speech to catch emotions text-based models miss

Block 发布 AI 语音模型 Dessa,号称首个完全基于人类语音、不含任何文本数据从零训练的模型,直接以音频处理语音,可实时捕捉情绪、讽刺、笑声和停顿。该模型支持全双工交流,可在 MacBook Pro M4 Max 本地运行,训练使用单台 NVIDIA GB200 NVL72。Dessa 尚未公开,团队正招募 AI 安全研究者测试其意外风险。

Hacker News front page

How Block Blast harvests its players for ad money

Ayo Omolewa 的技术拆解发现,Block Blast 从服务器下载规则,根据广告价格、观看记录和玩家分数请求无法完成的方块组合。测试中的规则要求最近 3 次激励广告均完成、待播放广告价值达到上次的 3.5 倍,且分数超过历史纪录 1,000 分;iPhone 实测还捕捉到另一条规则触发无解发牌。

The Verge · AI

Jack Dorsey’s Block made an AI model that talks to you like an awkward new friend

Jack Dorsey 的 Block 宣布语音模型 Dessa,仅用人类对话从零训练,演示交流大体自然,也出现认错人和突然跑题。团队称它是直接处理语音输入与输出的模型,所有知识均来自人类语音,可在 MacBook Pro M4 Max 或更高配置设备上本地运行。团队希望改善人与 AI 的理解及语音情绪识别相关安全问题,将向一小群 AI 安全研究人员提供早期访问权限。

Computing Life · Share · Yage

从 iSH 到 Open Minis:手机 Shell 已有多年,智能体运行时带来了什么

Open Minis 在 iSH 的 Linux 用户态环境上加入智能体运行时,让模型串联本地脚本、内置浏览器与 iOS 原生接口。它没有运行独立的 Linux 内核,受 iOS 限制采用解释执行;作者在 iPhone 上观察到部分工具调用失败,后台挂起、内存管理和授权也限制了任务执行。文章提出将手机的现场感知与长期智能体的记忆规划分层连接,通过主动推送和按需拉取接续跨端任务,而非仅同步技能文档。

Oct 23Friday

Hacker News front page

Show HN: Making a smart TV dumb and then smart

开发者发布 R2D2 for webOS 项目,为 webOS 电视提供本地主页、私有远程浏览器和按键说话语音助手,由一台 aarch64 Linux 主机运行 Chromium 与语音模型,电视仅作显示、应用和遥控端。该项目需电视 root 权限并修改厂商系统服务,仅针对特定固件版本验证,Home、浏览器和遥控语音已端到端跑通,其他电视与固件组合尚未验证。

Hacker News front page

Show HN: OnKey – I'm a bad singer, so I built an app to show where I'm off-key

Daniel Murrell 推出 OnKey,一款免费开源的 Mac 练唱工具,可学习歌曲旋律、实时显示演唱音高并逐句评分。歌曲播放结束后,它在本地分离主唱人声,用 CREPE 追踪音高、GAME 建模音符;需要 Apple Silicon(M1 或更新)及 macOS 14.2 或更新版本。

Oct 22Thursday

Hacker News front page

WorldTUI: I trained a 360M model to pretend to be a terminal

WorldTUI 基于 Qwen2.5-Coder-0.5B 微调,用 360M 参数模型根据按键预测终端屏幕变化,背后没有真实 shell 或文件系统。训练使用 336k 段真实终端会话,并通过错误修复训练改善一致性,通过并行生成屏幕行和顺序验证加速推理。

Oct 21Wednesday

AI HOT · Products

Perplexity Computer now runs fully locally on Mac, no cloud inference needed

Perplexity CEO Aravind Srinivas says Perplexity Computer can now run entirely on a Mac, with no cloud inference or WiFi. Both the model and the agent run on the machine, and it can operate native Mac apps and local files, so sensitive data doesn't have to leave the device. The feature is available now in the Perplexity Mac app.

Why it matters: Running the model and agent on the Mac itself means sensitive data stays local while it works with native apps and files.

Oct 20Tuesday

Computing Life · Share · Yage

员工用 AI 被辞退获法院支持、Anthropic 电视盒推理实验与 Character.ai 破产资产出售分析

北京朝阳法院判保险公司辞退使用第三方 AI 的员工违法,Anthropic 用电视盒合跑模型,Character.ai 申请破产保护。法院指出企业未明文禁用第三方 AI,也未举证文件涉密或实际损害;Anthropic 用 15 台约 55 美元的电视盒运行 Gemma4-26B-A4B,速度为 0.39 token/秒,研究尚未实现完整的失控自我复制。

TechCrunch · AI

Nvidia launches new open models to run agents in the enterprise and on PCs

Nvidia 周四发布 Nemotron 3.5 Super VL 和两款轻量 Bolt 开放权重模型,分别面向企业智能体与本地电脑部署。Super VL 拥有 120-billion 参数,支持组织在自有数据上后训练并部署到自有基础设施;Nemotron 3.5 Bolt 74B-A9B 24GB 面向本地多步骤推理、工具调用和网页浏览,Bolt E4B 面向对话智能体。

Oct 19Monday

Oct 18Sunday

Oct 16Friday

AI HOT · Products

NVIDIA RTX Spark laptops go on sale today, up to 128GB unified memory

NVIDIA's RTX Spark laptops went on sale worldwide today, with up to 128GB of unified memory and 1 petaflop of FP4 AI performance, starting at $2,599. The machines pair a Blackwell RTX GPU with a Grace CPU and CUDA, run large language models and AI agents locally, and add agent permission controls plus inference serving over a home network.

Why it matters: RTX Spark puts local model running, coding and testing inside a Windows laptop, cutting how much the dev loop depends on cloud test environments.

The Verge · AI

Meta and Apple team up to run Muse AI models better on Mac

Meta released Muse Spark Open with open weights and worked with Apple to run it through the latest MLX framework on two Apple M5 Ultra Mac Studios. Meta says the model is natively multimodal, has 816 billion total parameters, and handles long-horizon tasks, complex software engineering and multi-step reasoning. The two companies worked together for months, and Apple published a technical article covering MLX, Muse Spark Open and two other open models.

Why it matters: A months-long Meta and Apple collaboration lands as model and framework tuning that runs a large open-weight model on two desktop machines.

Hacker News front page

Meta releases Muse Spark Open weights, runs locally on two Mac Studios

Meta released the Muse Spark Open model weights with native multimodal support, 816B total parameters, 42B active parameters and a 1M-token context window, aimed at coding and agent tasks. The weights run locally on two Mac Studios.

Why it matters: Meta's open weights and two-Mac-Studio local run put coding and multimodal agent workflows on the desktop.

AI HOT · Models

Black Forest Labs releases FLUX 3 Image Klein, a 5.6B open-weight image model

Black Forest Labs released FLUX 3 Image Klein, a 5.6B open-weight model that handles text-to-image, image editing and multi-reference generation with up to 10 reference images. INT8 quantization lets it run on devices with as little as 8GB of VRAM, lowering the hardware bar for local image generation and editing.

Why it matters: FLUX 3 Image Klein runs on 8GB of VRAM with INT8 quantization, which lowers the hardware bar for local image generation and editing.

AI HOT · Models

NVIDIA releases open Nemotron 3.5 Super VL and two Bolt edge models

NVIDIA released Nemotron 3.5 Super VL and two Bolt edge models, covering agent deployment from enterprise servers down to local devices. Super VL is a 120-billion-parameter multimodal model with open weights, datasets and training recipe; it reads images, web pages and PDFs, and can be post-trained on an organization's own data.

Why it matters: NVIDIA's open models for servers and edge devices let organizations pick a deployment tier by hardware and tune agent tasks on their own data.