Show HN: WhatsApp MCP running free on Cloudflare
WhatsApp MCP 可部署在用户自己的 Cloudflare 账户,让 Claude Code、Claude Desktop 等 MCP 客户端读取和搜索 WhatsApp 聊天,但不能发送消息。
WhatsApp MCP 可部署在用户自己的 Cloudflare 账户,让 Claude Code、Claude Desktop 等 MCP 客户端读取和搜索 WhatsApp 聊天,但不能发送消息。
Netflix 发布新一代 VMAF 感知视频质量评估模型 v1,旧版 v0 模型文档继续保留,官方表示将持续根据用户反馈改进模型。VMAF 是曾获艾美奖的多方法融合视频质量算法,软件包包含 C 库 libvmaf 与 Python 封装库,并集成 PSNR、SSIM、MS-SSIM、CIEDE2000 等指标。
Browser Use 讲解浏览器智能体如何通过模型、运行框架和浏览器协作,循环观察页面、选择动作、执行并检查结果。任务有可用 API 时优先使用 API,流程可预测时使用脚本,需要灵活判断时使用智能体。智能体存在结果不一致、网站拦截、提示词注入和成本波动等限制,应设置运行上限,并核验任务是否实际完成。
大模型 API 的 ZDR 承诺有适用范围,文章梳理了应用状态、反滥用监控、缓存及法律保全可能保留数据的路径。文章结合 OpenAI、Anthropic 的政策说明,区分不用于训练与零数据保留,并强调部分端点和高级功能可能不受 ZDR 覆盖。工程应对包括检查调用链与第三方处理者、在数据出网前脱敏,以及为供应商切换或本地模型部署保留退路。
卡兹克实测华硕 ProArt 创16 RTX Spark 笔记本,128G 统一内存+1T 版本京东售价 37999 元,开售几分钟内被抢空。他受邀参加英伟达 RTX Spark 全球开售活动,并深度使用一周。
Codex Desktop 用户报告,Crashpad 转储持续积累至 ~323 GB、~4.04 million 个文件,占用 926 GB 磁盘的约 35%。
Mihai 提出将 AI 代码审查、自动化测试、安全扫描与人工复核结合,验证 AI 编码智能体生成的代码。文章比较 RoboRev、Claude Code Review、Codex、GitHub Copilot 和 Cursor Bugbot 的审查流程,并给出按提交、PR、部署及生产阶段分配检查的 GitHub Actions 示例,明确示例不能直接作为完整 CI 配置使用。
SGLang 推出 Low-Latency Mode,通过插件框架将专用 kernel 引入 SGLang,在 NVIDIA 和 AMD GPU 上显著降低推理延迟。
Neuralwatt analyzed cloud traffic and experimental requests and found that an agent harness editing an existing prompt breaks the cache, adding recomputation and energy use. With the same model and similar input and output lengths, requests that mostly miss the cache use about twice the energy of cache-hit requests, and about 2.5 times for GLM-5.3. When context passes 128k tokens, one broad re-read costs prefill compute roughly equal to 600 ordinary turns.
Why it matters: Neuralwatt traces cache misses request by request and frames stable prompt prefixes and appended state changes as ways to cut an agent's repeated computation.
China's commerce and science ministries released a draft revision of the technology export catalog that would put 23 key AI technologies under export control, covering large models, agents, compute clusters and data processing. A model's core technology is caught if it reaches 10^26 FLOPs of training compute, has at least 2 trillion parameters, or ranks in the global top 10. Models trained on China's "important data" and serving users from that training are also restricted.
Why it matters: China plans licensing thresholds for core large-model technology, with different procedures for overseas local servers and overseas cloud deployments, which directly shapes how models can be deployed abroad.
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 Commerce Ministry and Science and Technology Ministry issued Announcement No. 48 of 2026, proposing to add 23 key AI technologies to China's catalog of technologies banned or restricted from export, and opened the draft to public comment. Interim measures require registration before export. The proposed scope covers large models, agents, compute clusters and information processing. Technologies already in the public domain through open source or academic papers would not be included.
Why it matters: The draft sets where restricted large-model overseas deployment and offshore API calls fall, and the line between technology transfer and service provision shapes compliance for companies going abroad.
SGLang shipped Day-0 support for Muse Spark Open, an 816B open-weight multimodal model. The release adds sequence parallelism, NVFP4 MegaMoE, multimodal processing optimizations and DFlash V2 speculative decoding. Across seven concurrency levels, combining sequence parallelism with MegaMoE for long-prompt prefill cut p90 time to first token by 28.3% to 41.9%.
Why it matters: It pairs sequence parallelism with MegaMoE for long-prompt prefill and reports p90 time-to-first-token cuts of 28.3% to 41.9% across seven concurrency levels.
Meta released Muse Spark Open, an open-weight model with 816B total parameters, 42B active parameters and a 1M-token context window. The weights are available on Hugging Face under the Muse Model License.
Why it matters: Meta ships auto-tunable cross-hardware kernels and post-training support with the model, cutting the cost of building and maintaining custom kernels.
Meta used AI agents in Forum's planning, implementation, testing and verification, and a small team reached parity with Facebook Groups' core features in 8 weeks. The team supported agent work with readable app state, automated verification and build isolation. Engineers still joined some stages to iterate, and signed off on every shipped result.
Why it matters: Meta wired automated tests, UI checks and verification records into the agent workflow, so engineers review code against checkable results and keep the release sign-off.
Meta released the Muse Spark Open weights, and Baseten is serving the model the same day through Model APIs and Dedicated Inference, aimed at coding agent workflows. Baseten's NVFP4 deployment cuts the weight footprint from about 1.6 TB to 540 GB, freeing VRAM for concurrent agent sessions.
Why it matters: Baseten's NVFP4 deployment shrinks Muse Spark Open from roughly 1.6 TB to 540 GB, freeing VRAM for concurrent agent sessions.
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.
vLLM worked with Meta Superintelligence Labs to add day-one inference and LoRA RL support for Muse Spark Open. The open-weight MoE model targets long-horizon agentic coding workflows, with 816B total parameters, 42B activated per token, a 1M context window, and image, video and audio input. vLLM runs it across several hardware platforms and offers prebuilt images, so deployers can serve the model without waiting for the usual package merge.
Why it matters: Deployers can serve Muse Spark Open on day one through vLLM's prebuilt images instead of waiting for a standard package release.
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.
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.
Black Forest Labs released FLUX.3 Image Klein with same-day Diffusers support. One model handles text-to-image, multi-reference image generation and bounding-box layout control. Normalized bounding boxes set where generated regions go and where reference-image content moves, so layout instructions hold across different image sizes.
Why it matters: FLUX.3 lets you place generated regions and reference-image content with normalized bounding boxes that stay consistent across image sizes.
Modal closed a $750M Series D led by Accel at a $15.75B valuation, with annualized revenue run rate past $1B. Usage grew more than 15x year over year; the headline says usage and valuation each tripled in five months. Sandboxes, launched in 2023, have reached a $300M annualized run rate, with nearly 3 billion sandboxes started and support for a million concurrent.
Why it matters: Modal's sandbox business hit a $300M annualized run rate, and background agents, reinforcement learning and coding platforms are the concrete sources of that infrastructure demand.
AI-DLC plugs AI coding assistants into a structured, verifiable software delivery process through a tool-agnostic core. It supports Claude Code, Kiro CLI, Kiro IDE, Codex CLI, Cursor, opencode and GitHub Copilot.
Why it matters: AI-DLC ties requirements, implementation, testing and operations into one flow with approval gates and an audit trail, so decisions stay traceable across coding tools.
OpenCode 2.0 is a major release that connects the TUI, web interface and desktop app to a single local background service that runs on its own, and reworks the API. Code Mode turns MCP tools into functions scripts can compose by default, background subagents and async shell commands are new, TUI plugins are now stable, and an @opencode/sdk is still in development.
Why it matters: OpenCode 2.0 puts several clients on one backend service and moves subagents and long-running commands into the background, so the main agent stops blocking while it waits.
Anthropic 自建源码版本控制系统 Scion 替代 Git,由7人团队在5个月内完成迁移,以应对代码变更量的快速增长。每日合并的代码变更从2026年5月约2,500次增至9月超过30,000次,CI 每天约执行2,500万次代码检出,按朴素方式获取将产生超过600 PB的Git流量。
In its bankruptcy filing, Character.ai said high operating expenses and legal fees drove it under, and that the business was not sustainable even without the legal problems. Monthly compute spend ran at $10 million in 2025, while total revenue including ads peaked at just $5.3 million a month. As of August 2026, after cost cuts, it was still losing about $8-9 million a month.
Why it matters: The filing lays out the gap between Character.ai's compute spend and revenue, and shows losses continued after cheaper inference, layoffs and ads.
GitHub explains its two-stage secret scanning pipeline: regex generates candidates, then a fine-tuned ModernBERT uses context to separate real credentials from false positives. Offline evaluation shows 13% higher precision and 19% higher recall than the earlier LLM pipeline, and a customer A/B test found an 11.5% rise in alert acceptance rate weighted by repository owner. Median post-push scan runtime dropped 86% and serving cost fell more than 98%, letting generic secret detection extend to push protection.
Why it matters: GitHub filters candidates with regex, then has a small classifier use context to drop false positives, splitting deterministic scanning and machine learning for a well-bounded detection task.
Anthropic spent five months building Scion, its own version control system, and fully migrated its monorepo off git. Daily merged code changes rose from about 2,500 in May to more than 30,000 in September.
Why it matters: Anthropic moved version control compute to the server and used on-demand fetches and tiered caching to cut the cost of agent tasks keeping local repos.
A ChatGPT subscription can reach external dev tools three ways: an official API key, Sign in with ChatGPT, and reusing the Codex client identity. The article compares billing and usage limits for each.
Why it matters: It lays out how the three paths differ on billing, feature limits and protocol guarantees when wiring a ChatGPT subscription into dev tools.
Anthropic's cost guide says to compare models by cost per successful task, and to test effort settings before deciding whether a multi-model tier is cheaper. Switching models mid-conversation can break prompt caching, while changing effort does not. On SWE-bench Pro, running low effort first and rerunning failures at high effort scored 97.0% at $0.17 per task; running high effort throughout scored 95.3% at $0.29.
Why it matters: Anthropic sets single-model low-effort cost per task as the baseline for judging tiered setups, giving routing architectures a concrete cost benchmark.
Vapi built a review bot called Pedantic Review Bot that now reviews every PR engineers open. Average merge time fell from over 130 hours to 65. Before flagging an issue, the bot can trace calls, run tests or write a reproduction. The analysis job runs without Git credentials, and a separate job validates results and posts comments. The bot cannot approve or push code.
Why it matters: Vapi splits analysis from comment posting and requires tests or reproductions before reporting issues, a permission-isolation pattern for teams building their own review bots.
Anthropic's experiments show small AI models can complete several steps of self-replication, though a full runaway self-replicating system has not been achieved. The study ran Gemma4-26B-A4B on 15 Android TV boxes at 0.39 tokens per second, and Qwen3.5-9B deployed inference on 8 of 9 hardware types tested.
Why it matters: Anthropic tested inference deployment and end-to-end replication separately, drawing a clear line between small models adapting to unfamiliar hardware and completing the whole replication chain.
The design direction of an AI-generated interface depends on external rules. The article groups the base model, rule documents, component libraries and compilers into what it calls the interface generation core. It uses Anthropic's frontend-design skill document as an example: the flow outputs color, typography, layout and design principles first, then writes frontend components. It contrasts this with OpenAI Intelligent UI, which relies on standardized components and streaming rendering.
Why it matters: Anthropic writes the interface design process into a skill document, fixing color, typography and layout before generating components, a method others can reuse to constrain model output to a brand direction.
Piotr Sarnacki reviewed and re-tested an AI rewrite of Campfire Once. He argues developers still have to read the code and pin down constraints, and that a single metric cannot settle which language is better.
Why it matters: Sarnacki went through the AI-rewritten Campfire Once code and found the throughput comparison skewed by differences in compatibility and queue design between the versions.
Yandex Cloud says its data center in Vladimir was damaged by a drone strike and is fully out of service, with no casualties. The cloud platform is in emergency mode: ru-central1-a and ru-central1-b are unavailable, and the company recommends disaster recovery plans on other platforms.
Why it matters: The strike took a Yandex Cloud data center offline and pushed the platform into emergency mode, with compute running in other availability zones but storage writes and some databases still affected.
Over three months a team used AI agents to rebuild the C++ source of a first-person shooter, covering 99% of functions, with 83% matching byte-for-byte. Early code ran but was semantically wrong, so the team switched to the original compiler and byte-match checks, refreshing instructions hourly and blocking agents from drifting off task or editing the verification scripts. The author estimates real usage at 600-700 billion tokens, calls the project done, and will keep the code private.
Why it matters: The team turned correctness into a machine-checkable acceptance test, so reviewers are not swayed by excuses written into code comments.
Microsoft Execution Containers (MXC) is now generally available. It enforces file, network and UI access policies through containers to limit what AI agent workloads can do. Developers configure it with a unified JSON setup and multi-language SDKs across Windows, macOS and Linux isolation backends, and the Windows process container can log resource access to help tune policies.
Why it matters: MXC moves agent permissions into the runtime itself, and its learning and permissive modes record resource access so teams can tighten least-privilege policies with real data.
Prime Intellect released a Rust rewrite of Prime Agent, migrated end to end in two weeks by more than 2,000 agents. The release adds native Windows support in beta and a homebrew install.
Why it matters: Planning, implementation, review and verification went to separate agents, with differential testing gating the migration instead of heavy human review.
Google has open-sourced AQuA (Ambient Quality Agent), which analyzes production sessions alongside a project on Google Cloud and pairs them with source snapshots taken at deploy time to diagnose agent failures.
Why it matters: AQuA clusters failures from production sessions and ties them to deploy-time source snapshots, giving traceable evidence with per-cluster checks and code line references for agent debugging.
A reading of the GPT-6 guide lays out how to build long-running agents, shifting the focus from homegrown memory management to collaboration and authorization while a task runs. Drawing on OpenAI documentation, it covers compaction and persistent reasoning, mid-turn steering, asynchronous tool calls and multi-agent delegation, plus steps for model choice, caching, skill rules and human approval. It also flags limits: correction instructions lost on disconnect, mutually exclusive API parameters, and encrypted compaction that is hard to audit.
Why it matters: It splits long-task engineering into managed memory and collaboration control, and assigns clear roles to async execution, mid-run correction and human approval.