Tag
A hobbyist built a distributed inference system chaining two gaming PCs, an M4 Pro MacBook, and two iPhone 17 Pro Maxes to run Qwen 3.8 Flash Next (a 48-layer MoE model with 512 experts), achieving 37-39 tok/s at 60k context via pipelined speculative decoding across the device chain.
nanoMuse 是一个开源个人智能体（GPL-3.0），可在用户拥有的每台设备上运行，通过共享中继实现跨设备对话，所有操作经 Sentinel 审核，记忆以用户可读的文件形式存储，模型由用户自选。论文还分析了 Meta Muse 的构建方式，并规划了带来源的记忆、手部操作评估套件及开放模型的路线图。
A user connected Hermes gateways across multiple machines to centralize control of AI agents and bots in a single desktop interface, creating a unified fleet management system.
X-Stream introduces the first benchmark for multi-stream video understanding, evaluating MLLMs as multiplexers across multiple concurrent streams. The study reveals that current MLLMs achieve only about 50% accuracy, exposing significant limitations in handling multiple streams.
OpenAI Codex adds a remote control feature, allowing you to connect and operate multiple computers via ChatGPT, and share context and files, greatly improving cross-device collaboration efficiency.
A user shares his experience setting up Codex across multiple devices (MacBook, Mac mini, phone) with SSH, creating an always-accessible AI coding assistant that feels like a persistent presence rather than being tied to a single computer.