Tom's Hardware · 1h # ☆

gputaskman open-source tool shows what's using VRAM

gputaskman is a new open-source GPU task manager from Microsoft VP Scott Hanselman that reveals which processes are consuming your VRAM.

Reddit · 2h # ☆

Developer rebuilds IDM as open-source download manager with Claude Code

Mister Download Manager is a free, open-source download manager written in Rust, built with Claude Code as an alternative to IDM; version 0.5 shipped.

Hacker News · 2h · 2 sources # ☆

llama-modchip runs 284B DeepSeek model on 64GB RAM without GPU

Why it matters: Shows large MoE models can run on commodity CPU RAM, lowering local inference barriers.

An open project called llama-modchip runs a 284B-parameter DeepSeek model on a 64GB CPU-only system, showing off-device inference for large models without a GPU.

Bloomberg · 2h · 4 sources # ☆

Apple explores turning Siri into ChatGPT Dots-like AI agent

Why it matters: Signals Apple's planned pivot from a voice assistant to an autonomous on-device agent across apps.

Bloomberg reports Apple CEO is pressing its software chief to pursue a more ambitious Siri AI strategy, exploring an agent that works like ChatGPT Dots. Some AI leadership departures are rumored, and broader app control may wait until spring 2027.

Linuxiac · 2h # ☆

Zabbix 8.0 LTS adds OpenTelemetry, ClickHouse and AI features

Zabbix released version 8.0 LTS of its open-source monitoring platform, adding OpenTelemetry support, ClickHouse storage and AI integrations for infrastructure observability.

SecurityWeek · 3h # ☆

Brava Security emerges from stealth with $15M

Brava Security emerged from stealth with $15 million in funding. Its multi-agent AI platform generates the telemetry an attack would produce to help enterprises fix defense gaps.

MarkTechPost · 4h # ☆

Google releases Elephant-Goldfish plugin for spec-first coding

Google released the Elephant-Goldfish model plugin, which makes coding agents draft specifications and run five zero-context review stages before writing any code.

Reddit · 5h # ☆

Reddit user shares ComfyUI agent skill for readable workflows

An agent skill for ComfyUI that rewrites messy node workflows into more readable, understandable layouts, shared by a ComfyUI tutorial site operator.

Informat.ro · 5h · 3 sources # ☆

Anthropic reportedly lowers valuation estimate ahead of Nasdaq listing

Why it matters: A trimmed Anthropic valuation would set a key benchmark for AI lab IPO pricing.

Anthropic has lowered its estimated valuation ahead of a planned Nasdaq IPO amid market turbulence and AI investment concerns. Analysts say the expected IPO price demands near-perfect execution; Broadcom's AI growth case is also tied to its S-1.

NoMac gives AI coding agents a cloud Mac billed per minute

NoMac provides Claude Code, Codex or Cursor its own cloud Mac with Apple silicon, ready in about 30 seconds, so the agent can build, run on an iPhone Simulator, and ship to TestFlight and the App Store. Billed per minute from prepaid credit, no subscription.

Agent Clan controls Claude Code and Codex agents with gestures

Agent Clan turns Claude Code and Codex agents into a 3D team on screen controlled by hand gestures and voice: make a fist to summon them, pinch to select, pull right or left to allow or deny permissions, and circle screen content to hand it over.

Sage2 launches APIs for structured data from live talk

Sage2 by Levanto Labs gives developers APIs to extract structured data such as dates, amounts and durations from natural language and make streaming decisions as conversations evolve, tracking changing user intent.

AXL launches GTM copilot for AI product founders

AXL is a go-to-market copilot for AI founders combining pages, quizzes, webinars, courses, email, chat, CRM and AI calls, with 1,200+ actions via MCP and a model trained on 1M deals that scores leads using 300+ signals.

Memorias launches AI biographer that turns stories into a book

Memorias turns spoken life stories into a finished memoir: an AI biographer asks follow-up questions, transcribes sessions, maps people, places and years into a timeline and assembles chapters checked against what was said. Exports EPUB, PDF or DOCX.

Otto launches voice AI CRM for field sales teams

Otto is a voice AI customer relationship management tool aimed at field sales teams, letting reps log and manage sales activity by voice rather than manual entry.

Howl.Link · 8h # ☆

Shizuku AI releases Japanese full-duplex speech model

Shizuku-Duplex is a Japanese speech model that handles two-way conversation in real time, letting a voice agent listen and respond without taking strict turns. The release targets natural, interruptible Japanese dialogue.

DD News · 11h · 2 sources # ☆

Mistral launches Large 4 model with open weights

Why it matters: A new open-weights frontier-class model from a major European lab intensifies the open vs closed race.

Mistral released Mistral Large 4, an open-weights model it says outperforms some Chinese rivals, with published license, pricing and benchmark results.

Microsoft Source · 12h # ☆

Elo and Microsoft launch AI agent to automate payments in Excel

Why it matters: Pushes agentic payments into the spreadsheet, a common corporate finance tool.

Elo and Microsoft announced an AI agent solution that automates corporate payments within Excel, according to Microsoft's newsroom.

RuntimeWire · 14h # ☆

Cubicle queues blocked coding agents at your desk

Cubicle is a free open-source tool that renders Claude Code, Codex, Gemini CLI and Paperclip sessions as characters, with blocked agents waiting at a desk.

Reddit · 14h # ☆

Open-source MCP server lets AI agents play PC games

An open-source MCP server lets agents like Claude or Codex drive Windows games in the background, including driving, walking and flying in GTA, without taking over the desktop.

Reddit · 18h # ☆

Developer builds Mac app of desktop pets run by local Gemma 3 1B

A Mac app puts nine animated desktop pets on your windows, driven by gemma3:1b through Ollama so they nap, steal things and ignore you. Runs locally; a Reddit side project.

Yahoo Finance · 19h # ☆

Concentrix launches AI migration tool

Concentrix launched an AI migration tool, and coverage examines what the launch means for the company's shareholders.

Reddit · 21h # ☆

Developer builds Bodylang open-source AI anatomy explorer

Bodylang is an open-source interactive anatomy tool that lets users explore the human body in 3D with AI, built by a developer after being impressed by Opus 5.5's 3D modeling.

Reddit · 21h # ☆

Observer app runs local models to monitor phone camera

Observer is an open-source, fully local iOS and Android agent that uses on-device models to watch the camera feed and send alerts, such as when someone approaches a car or a bird lands on a nest.

Tekedia · 22h · 2 sources # ☆

Google adds prompt-based revisions to Gemini Omni video model

Why it matters: Prompt-based editing in a major Gemini video model pushes iterative AI video workflows.

Google advances AI video generation with Gemini Omni 1.2 Flash, adding conversational prompt-based revisions to enterprise video models. Google also committed $17 billion to a Wyoming data center.

Cult of Mac · 22h # ☆

GhostShield VPN adds real-time AI threat protection

GhostShield VPN Premium adds an AI Threat Analyzer plus ChaCha20 encryption, unlimited bandwidth and support for 10 devices.

RuntimeWire · 23h # ☆

Devvm runs coding agents in per-project local VMs

Devvm is an open-source tool by CJ Reynolds that gives each project folder its own Lima VM for Claude Code, Codex or OpenCode, with allowlisted networking by default.

Reddit · 23h # ☆

Developer releases open end-of-turn detector for English and Indic voice agents

An open end-of-turn detection model fine-tuned from Qwen2.5 handles English plus eight Indic languages so voice agents stop interrupting callers mid-sentence, where verb-final grammar confuses plain VAD.

The Verge · 23h · 2 sources # ☆

NYT Cooking launches AI recipe modifier assistant

The New York Times' Cooking app has added an AI-powered recipe modifier that adjusts dishes to taste, diet or ingredients. Reviewers found it helpful for weeknight meals, though results varied with baking.

Reddit · 23h · 2 sources # ☆

Microsoft engineer builds tool showing GPU memory usage

A Microsoft engineer built a tool that shows what is really using your GPU, and new video memory monitoring is available for Nvidia and AMD cards. Useful for AI workloads that fill VRAM.

Dugout Philippines · 1d # ☆

Kravos AI launches public beta of enterprise AI agent infrastructure

Kravos AI opened the public beta of its enterprise-grade infrastructure platform for building, running and governing AI agents.

Reddit · 1d # ☆

Valet launches open-source dashboard for managing homelabs with local LLMs

Valet is an open-source dashboard that uses local LLMs to help manage homelab services, keeping everything private. A chat agent works with any local or hosted model; a live demo is available.

Reddit · 1d # ☆

Terrarium runs AI agents in local Linux VMs with explicit permissions

Terrarium is an open-source tool that runs AI agents in persistent local Linux VMs called boxes and grants each explicit permissions, so agents can work against local models without full machine access.

ET Enterprise AI · 1d # ☆

Amazon rolls out AI shopping assistant over text messages

Amazon launched an AI-powered shopping assistant in the US that lets customers search, ask questions and shop by texting, extending its AI commerce tools to messaging.

ShiftDelete.Net · 1d # ☆

GitHub Copilot arrives in Xcode in public preview with 27.3

GitHub Copilot is now available inside Apple's Xcode in public preview with version 27.3, letting developers use the coding assistant directly in Apple's IDE.

Reddit · 1d # ☆

Two-step LoRA for p-h3-image image model released

A Photon-style image model adapter, p-h3-image 2-step LoRA, lets users generate images in two sampling steps at lower compute; weights posted on Hugging Face.

Techgenyz · 1d # ☆

Microsoft Foundry adds five AI video models via fal partnership

Microsoft Foundry now offers five AI video generation models through a partnership with fal.

Pasquale Pillitteri · 1d # ☆

Qwen3.5 9B uncensored merges DeepSeek-V4-Pro reasoning

An uncensored Qwen3.5 9B model arrives with DeepSeek-V4-Pro reasoning distilled in, reportedly refusing only 3 of 100 test prompts.

MarkTechPost · 1d # ☆

Buoyant releases Kinference inference control plane for Kubernetes and vLLM

Why it matters: An open-source control plane could simplify running and scaling self-hosted LLM inference at lower latency.

Kinference is an Apache 2.0 control plane that manages vLLM inference on Kubernetes, adding declarative routing, KV-cache-aware load balancing and per-tenant metrics. Buoyant says its benchmark cut median TTFT by 95%.

Quest launches personal AI advisor for intentional living

Quest is a personal AI advisor that learns a user's personality, values and ambitions to help them reflect and make decisions, rather than just completing tasks.

RS Web Solutions · 2h # ☆

Zorio X6 smart band launches at 3,999 with free AI app

The Zorio X6 smart band launches at 3,999 and includes a free AI companion app for health tracking.

Phoronix · 3h # ☆

Linux 7.5 adds more AMD Ryzen AI NPU AIE4 enablement

Phoronix reports Linux 7.5 will include additional enablement for AMD's Ryzen AI NPUs based on the AIE4 architecture, improving kernel support for on-device AI acceleration.

Wccftech · 4h # ☆

DFRobot DX-M1 M.2 AI accelerator offers 25 TOPS at 5W for $139

DFRobot's DX-M1 is an AI accelerator that fits in an M.2 slot, delivering 25 TOPS at 5W for $139, though it is not suited to running heavy LLMs.

Business Insider Africa · 8h · 3 sources # ☆

Ex-Nvidia engineer brings GPU cloud to six African nations

Alexander Tsado, who helped build Nvidia's early cloud infrastructure, left the company to launch UduTech, a GPU cloud operating in six African countries.

Pandaily · 8h # ☆

Moore Threads updates AICUBE with smart-home control and NAS features

MTT AIOS 1.6.1 adds Matter/ZigBee smart-home control, Olares file management, semantic photo search and a desktop-pet mode to Moore Threads' AICUBE device.

Pandaily · 13h # ☆

X-Humanoid releases Hi-WBC whole-body control model

Why it matters: Whole-body control combining motion priors and force sensing is core to making humanoids useful.

X-Humanoid released Hi-WBC, a whole-body control model for its Tiangong humanoid robots that blends human motion references with force sensing and terrain perception; demos showed sports events and room tidying.

The Indian Express · 1d · 3 sources # ☆

CoreWeave launches Physical AI Field Engineering service

CoreWeave announced Physical AI Field Engineering, a service aimed at industrial physical AI deployments. Coverage ties it to a broader 'physical AI' frontier for industrial applications.

iThinkDifferent · 1d · 3 sources # ☆

Siri AI memory graph in iOS 27.3 beta learns user routines

Why it matters: Apple's personal context layer could reshape how Siri competes with assistant rivals.

Apple's iOS 27.3 beta adds a personal knowledge graph that lets Siri learn your routines and life details across iPhone and Mac, per early reports.

AiMOGA's Mobo home companion robot opens global pre-orders

Mobo, a home companion robot from AiMOGA, debuted globally and opened pre-orders as AiMOGA showcased its robot family at the 2026 OMODA & JAECOO International User Summit. Ynetnews rounds up five home robots pitched at chores.

Boston Dynamics built Stretch Gen 2 for reliable logistics work

Boston Dynamics' new Stretch Gen 2 warehouse robot took two and a half years to develop and targets more reliable, consistent logistics performance, per its logistics VP.

Mint · 1d # ☆

Motorola Signature 27 launches with Gemini AI and 200MP camera

Motorola launched the Signature 27 smartphone with a 200MP camera, Gemini AI features, a 7,000mAh battery and GrapheneOS support.

AWS Machine Learning · 1d · 2 sources # ☆

AWS SageMaker Inference adds Nvidia MIG GPU partitioning

Why it matters: GPU partitioning lets cloud users pack more models per accelerator and cut inference cost.

Amazon SageMaker Inference now supports Nvidia Multi-Instance GPU, splitting one GPU into isolated partitions so users can run separate models or right-sized fractions of a newer accelerator per endpoint.

Gizmochina · 1d # ☆

DJI launches Matrice 5T industrial drone globally

Lightweight industrial drone ships globally with standard LiDAR, millimeter-wave radar, omnidirectional vision, 176x zoom and 59-minute flight time.

MarkTechPost · 1d # ☆

Zyphra releases ZYFA FP8 attention kernels for AMD MI355X

Why it matters: Gives AMD GPU users faster, lower-error attention kernels as an open alternative to AITER.

Zyphra open-sourced FP8 and BF16 prefill attention kernels for AMD Instinct MI355X that run 1.16-1.76x faster than AMD's AITER, with 62-78% lower error at head dimension 128.

Unite.AI · 1d # ☆

Baseten makes Serverless Loops generally available

Why it matters: Per-token post-training billing lowers the barrier to fine-tuning open models.

Baseten's Serverless Loops, a managed SDK for supervised fine-tuning and reinforcement learning of open models with per-token training billing, is now generally available and connects checkpoints to Dedicated Inference.