1
AI本地部署不如官方版的元凶找到了：734个依赖包，每一个都可能坑
推理软件栈的微小差异，就能改变输出token
2
I Built the World's First AI-to-AI Communication Language
AI 模型间互不理解的痛点，用一门全新编程语言来解：开源、本地运行，内置编译器和 50+ 张量操作。
🟢 I Built the World's First AI-to-AI Communication Language TL;DR: Создал Vireo — язык программирования для ИИ-моделей. Компилятор, интерпретатор, API…
3
Show HN: ChatOSS – A Codex alternative for Open Source AI built on Ollama
开源AI编程工作台，本地跑模型，AI代理写代码、管看板，一站式交付
ChatOSS is built on Ollama. If you use Ollama, ChatOSS local works out of the box. ChatOSS is a GUI desktop app that has multiple agentic coding apps …
4
5
Show HN: Dictata – Local Whisper dictation with LLM cleanup
本地语音转文字加AI润色，隐私安全还免费
Article URL: https://github.com/AntoineChatry/Dictata Comments URL: https://news.ycombinator.com/item?id=49306462 Points: 3 # Comments: 0
6
Show HN: A Linear agent that runs Claude Code sessions on your own machine
把 Linear 工单自动变成本地 Claude Code 会话，轻量实现、串行执行，适合自动化工作流玩家。
I keep everything in a local Obsidian knowledge base and run Claude Code against it all day. I also live in Linear. I wanted to ask the knowledge base…
7
Show HN: Visual knowledge graphs using Ollama and Embeddings
用Ollama和Embeddings本地构建可视化知识图谱，开源玩法值得一试。
Article URL: https://github.com/punnerud/Local_Knowledge_Graph Comments URL: https://news.ycombinator.com/item?id=49270102 Points: 1 # Comments: 0
8
Unsloth Desktop
本地跑AI模型还支持微调，隐私安全且零门槛，桌面端神器。
Run and train AI models locally on your desktop Discussion | Link
9
Show HN: ExpertCache – Run the Full GPT-OSS 120B (63GB) on a 16GB M1 Pro
缓存技术让63GB的GPT-OSS 120B在16GB M1 Pro上完整运行，本地推理大模型不再是梦想。
Article URL: https://github.com/amos-labs/expertcache Comments URL: https://news.ycombinator.com/item?id=49226502 Points: 1 # Comments: 0
10
ThonburianTTS — AI พูดภาษาไทย ฟรี รันในเครื่องได้ โคลนเสียงได้ โดยคนไทย
免费泰语TTS可本地运行、克隆声音，但研究称长文本生成有硬伤，泰国开发者项目值得关注。
ThonburianTTS — AI พูดภาษาไทย ฟรี รันในเครื่องได้ โคลนเสียงได้ โดยคนไทย โดย Nokka (นก-กา) | 8 สิงหาคม 2569 บทความนี้เขียนโดย AI (deepseek-v4-pro) ผ่าน…
11
Homebot: A Personal AI Agent for Conversational Home Assistance and Automation
家庭场景下的个人AI智能体方案：用自然对话搞定家居自动化与控制，多模态交互+任务规划新思路，AI产品人可重点关注。
arXiv:2608.02254v1 Announce Type: new Abstract: \texttt{Homebot} is a locally deployable AI agent for conversational household assistance and automati…
12
I gave local AI eyes
开源神器让本地AI实时看图，手机秒变AI眼睛，支持历史帧回放，部署简单
Article URL: https://github.com/zyads/eyes Comments URL: https://news.ycombinator.com/item?id=49147881 Points: 2 # Comments: 0
13
Testing LLM Concurrency on Consumer Hardware (RTX 5060)
RTX 5060 八张显存实测 19 款模型并发推理，最高提速 8.7 倍，小显存本地部署选手的硬核参考。
Article URL: https://ai.2it.onl/posts/concurrency-sweep/ Comments URL: https://news.ycombinator.com/item?id=49141833 Points: 2 # Comments: 0
14
Running a 35B LLM at 128K Context, Full Speed, on €870 of Used Hardware
870欧元二手硬件全速跑35B大模型，128K上下文本地部署的极限性价比方案。
Article URL: https://medium.com/ai-advances/running-a-35b-llm-at-128k-context-full-speed-on-870-of-used-hardware-no-cloud-required-c4f7629810b8 Commen…
15
Open-Source AI Platform Orbit
自托管AI平台，Docker一键部署，本地离线运行大模型，兼顾隐私与便捷
Article URL: https://github.com/schmitech/orbit Comments URL: https://news.ycombinator.com/item?id=49134457 Points: 2 # Comments: 0
16
Setup instructions for every major AI provider – US, Chinese, and local
一份覆盖中美主流及本地AI模型的接入指南，快速搞定API密钥与配置，开发者必收。
Article URL: https://big-agi.com/docs/connect-models Comments URL: https://news.ycombinator.com/item?id=49116394 Points: 2 # Comments: 1
17
Show HN: Run GLM-4.5-Air(110B)on a 16GBRAM consumer machine
16GB内存就能运行110B大模型，量化技术让消费级硬件也能玩转前沿AI。
Article URL: https://github.com/FedericoTs/quantprobe Comments URL: https://news.ycombinator.com/item?id=49028865 Points: 1 # Comments: 0
18
Familiar – Local AI Workspace with Chat, Notes, Wiki, and Automations
本地AI工作空间，集成聊天、笔记、维基和自动化，支持高效知识管理与任务流
Article URL: https://github.com/sixvolts/familiar Comments URL: https://news.ycombinator.com/item?id=48930009 Points: 3 # Comments: 0
19
Cognitive Defrag: A Local Sense Filter for Anti-Chaos Thinking (Zero-Data Edition)
用Python+aiogram打造本地文本去碎片化Telegram机器人，零数据泄露实现思维整理
The Manifesto: Against the "Cognitive Impasse" Let’s be honest. The tech world is currently drowning in poorly formulated requirements, chaotic task d…
20
Getting GLM 5.2 running on my slow computer
亲测GLM 5.2在慢电脑上的运行体验，能力逼近闭源模型但面临内存瓶颈，本地部署党必看
A few days ago I found myself trying out GLM 5.2 and was really positively impressed. The capabilities and security I was getting from this LLM are si…