Interfaces

Even Realities G2 review: Smart glasses, stupid controlsEven Realities makes one of the best smart glasses on the market, but controlling its growing feature set is harder than it should be.4 min

How the Xbox’s Disc-to-Digital Program WorksMicrosoft is letting players digitize their game discs for free, but don't sell your physical collection just yet.4 min

Disney Asked Character.AI to Go Bankrupt for a Get-Out-of-Lawsuits CardThe entertainment giant hopes to get trust and safety tech, among other IP, from a startup whose chatbots dozens of states allege encouraged minors to self-harm and kill their pare4 min

Tmux and Claude Code: Sessions, Send-Keys, and the CeilingHN score 164 min

openai-python v3.32.03.32.0 (2026-10-16) Features api: add cache settings and key options [2/2] ( #4087 ) ( 325da45 ) api: add prompt-based voice creation ( #4091 ) ( 2e39db7 ) api: add regional proces4 min

Your API Model Name Stayed the Same. The Model Didn’t.On Saturday, October 17, 2026, late in the afternoon Beijing time, every app calling deepseek-flash on DeepSeek's API started talking to a… Continue reading on Medium »4 min

Two Cheap Agent Models, Ten Days Apart: DeepSeek Flash 1017 vs Claude Haiku 5.5On October 7, Anthropic released Claude Haiku 5.5 at ten cents per million input tokens and fifty cents per million output tokens, and… Continue reading on Medium »4 min

openai-python v3.34.03.34.0 (2026-10-17) Features api: add degraded external storage status ( #4097 ) ( b216cb2 ) api: add Realtime Translation WebSocket sessions ( #4095 ) ( bfada50 )4 min

openai-python v3.35.03.35.0 (2026-10-17) Features api: add typed Live sideband events ( #4099 ) ( 8d5cbf0 ) api: add ultrafast service tier for Agents ( #4101 ) ( 100bf8f ) Bug Fixes realtime: preserve4 min

llama.cpp b11720ggml: support non-contiguous SUM (CPU, CUDA, Vulkan, WebGPU) ( #28851 ) ggml: support non-contiguous SUM (CPU, CUDA) tests: add non-contiguous SUM cases tests: SUM over non-contigu4 min

llama.cpp b11723hexagon: FLASH_ATTN_EXT improvements for longer context and SWA model…4 min

openai-python v3.35.13.35.1 (2026-10-18) Bug Fixes preserve beta Realtime byte reads on websockets 13 ( #4102 ) ( 5766349 )4 min

This Samsung Galaxy feature can protect you from untrustworthy appsSamsung's Auto Blocker protects your Galaxy phone from security threats, but it does have drawbacks. Here's what it does and when you might want to turn it off.4 min

How the Claude Code team does code reviewHow do you review code at the volume agents produce it? The Claude Code team shares its review process, from Claude verifying its own work to risk-tiered human approval, plus a che4 min

Meta’s new terms may allow it to train AI on news sites for freeMeta's new terms appear to give it permission to crawl and scrape websites "associated" with the businesses that use its platforms - including online publishers, according to a rep4 min

PyTorch viable/strict/1792135796[inductor] Let TritonKernel subclasses choose the generated driver se…4 min

llama.cpp b11680hexagon: tensor repacking rework and support for tiled and flat quantized and fp weights ( #30635 ) hex-repack: Remove need_repack hack and streamline the rules -- breaks GET_ROWS 4 min

llama.cpp b11682: Fix failing RocM tests (#30611)Co-authored-by: Karthik Vijayan kavijay@microsoft.com4 min

llama.cpp b11681graph : add precision handling for MELLUM architecture ( #30584 )4 min

llama.cpp b11678Limit max CUDA pinned host buffer size to 4 GB ( #30449 )4 min

Business accounts will consent to Meta scraping their websites Starting Oct. 30HN score 644 min

The Tesla Roadster 2 Is Here—and It FliesThe automaker unveiled its next luxury sports car at an event in Texas. It has thrusters on it that can send it airborne.4 min

llama.cpp b11672mtmd: lightonocr does not require longest edge to always be 1540, nat…4 min

Trump posts AI-edited photo to brag about White House restoration workPresident Donald Trump posted a series of photos on Truth Social showing off renovations to the White House - but one of them appears to be modified with AI. The picture, which com4 min

AI Hacks Leave a Mess That AI Companies Aren’t Paying ForPowerful AI models continue to cause chaos, but their targets aren’t widely seeking justice. Instead, they’re being encouraged to adopt those same tools.4 min

Code Review is now generally availableCatch more bugs before they reach production.4 min

Next.js v16.4.2What's Changed [lts-active] turbo-tasks snapshot consistency fixes ( #99644 , #99732 , #99649 ) by @lukesandberg in #100002 [lts-active] Turbopack: backport Pages Router dev module4 min

Modernizing perimeter security: Scalable perimeters with VPC Service ControlsAgents represent a fundamental shift in cloud operations. They can act on behalf of users or autonomously, make non-deterministic decisions, and request data across projects at unp4 min

Amazon DynamoDB Streams now supports timestamp-based stream processingAmazon DynamoDB Streams now supports timestamp-based stream processing, enabling you to begin reading stream records from a specific point in time. Previously, you could only start4 min

More on Anthropic's Scion ripping off MononokeHN score 654 min

Friends don’t let friends feed their data to AI agentsThere's a growing trend in the consumer tech world: gadgets designed to spy on everything and everyone around you, regardless of whether or not anyone else actually consented to it4 min

Dial Up the Cloud: NVIDIA DLSS 4.5 Controls Arrive on GeForce NOWDial in the next gaming session this GFN Thursday. New NVIDIA DLSS 4.5 controls give GeForce NOW members more choice over how supported games look and perform in the cloud, even wi4 min

How Windows is being optimized for Nvidia’s RTX Spark laptopsMicrosoft and Nvidia held a launch party for the first wave of RTX Spark laptops last week in San Francisco, where a new wave of AI PCs took center stage amid a packed crowd and a 4 min

NVIDIA's RTX Spark is good for Windows, even if you don't need local AINVIDIA's RTX Spark is meant for AI fiends, but it could end up helping Windows PCs in general4 min

llama.cpp b11647vulkan : copy strided f16 KV once on AMD so reads spread across memory channels ( #27703 ) vulkan : copy strided f16 KV once so reads spread across memory channels Assisted-by: Cla4 min

Bee unveils two AI note-taking devices, its first new products since being acquired by AmazonThe Eclipse and Frontier are cheap and colorful, but you'll need a Pro subscription to make the most of them.4 min

The Design Space of Tri-Modal Masked Diffusion ModelsDiscrete diffusion models have emerged as strong alternatives to autoregressive language models, with recent work initializing and finetuning a base unimodal model for bi-modal gen4 min

Six wild ideas for perpetuating your mindHave you ever thought about what makes you you? Science assures us it has everything to do with the three-pound organ between our ears, with its billions of neurons and trillions o4 min

One Token at a Time: How Autoregressive Models Work, Where They Break, and What Is ChangingWhy this matters Continue reading on Medium »4 min

Gemini 4 Argon vs GPT-6 Astra vs Claude Opus 5.5:Key Takeaways Continue reading on Medium »4 min

How to Use Muse Wonder API on GPTProto: API Keys, curl & Python Examples, Pricing and…Key Takeaways Continue reading on Medium »4 min

llama.cpp b11640vulkan: Use L2-friendly workgroup scheduling for cm2 mulmat ( #30215 )4 min

llama.cpp b11635ggml-cpu : keep the FP16->FP32 lookup table on x86 with F16C ( #30342 ) ( #30343 ) ggml-cpu : keep the FP16->FP32 lookup table on x86 with F16C ( #30342 ) #30270 routed both 4 min

llama.cpp b11636: sycl: IQ3 code reorder (#29107)sycl: persistent IQ3_S/IQ3_XXS weight reorder for SYCL decode and prefill Add persistent reordered weight layouts for IQ3_S and IQ3_XXS. Weights are reordered once at load time; th4 min

Fitbit Edge vs Fitbit Air: The differences beyond a screenGoogle's latest wearable seems like a Fitbit Air with a screen, but the two actually have less in common than you might think.4 min

Next.js v15.5.28This release contains security fixes for the following advisories: Critical: Unauthenticated Remote Code Execution in Image Optimization API when SVG files are used High: Denial of4 min

llama.cpp b11630ggml: fix conv_1d result layout for batched input ( #28738 ) ggml: fix conv_1d result layout for batched input The GEMM behind conv_1d folds the batch into its rows, so the product4 min

Next.js v16.5.0-canary.12Core Changes Upgrade from pnpm v10 to v12: #99827 Example Changes docs(examples): fix typo and update HTTP links to HTTPS in example READMEs: #99924 Misc Changes Turbopack: Avoid s4 min

McKinsey is making job applicants spar with a fake AI clientConsulting firm McKinsey & Company is going to ask some entry-level candidates to have a 10-minute conversation with an AI simulation of a client as part of the interview process, 4 min

llama.cpp b11627ggml : add note about mul_mat_id + disable mul_mat_id_dup tests ( #30501 )4 min

Next.js v16.5.0-canary.11Core Changes Upgrade React from b618bbb4-20261007 to 07dde64d-20261013 : #100194 Remove staticInfo.rsc: #100172 fix(analyze): use route summaries as the analyzer route index: #10004 min

The AI race is back onLast week, Anthropic and OpenAI looked like they had the AI race to themselves, pulling ahead with increasingly powerful models that left their rivals in the dust. Google seemed ho4 min

AI Came for Math. Now It’s Chasing the Rest of ScienceAI may soon be the default way to solve theoretical problems in economics, high-energy physics, and other fields, MIT researchers tell WIRED.4 min

Next.js v16.4.1This release contains security fixes for the following advisories: Critical: Unauthenticated Remote Code Execution in use cache functions returning typed arrays Unauthenticated Rem4 min

Predictable AI spend in BigQuery: Introducing cost management for generative AI functionsThe generative AI functions in BigQuery ( AI.GENERATE , AI.IF , AI.CLASSIFY , AI.SCORE , and more) bring large language models (LLMs) directly into your SQL, so a single query can 4 min

How Third Moment Research runs 10,000 agents for 36+ hoursWith GPT-6 Astra and GPT-6.1 Sol, Codex, and Daybreak for cybersecurity, this AI lab builds agents that focus for days and improve with use.4 min

Why a better chatbot isn’t what most teams need, and what a shared workplace for people and AI…Picture a Tuesday afternoon. A new client has just signed, and you ask your AI assistant to draft a welcome email. It writes a good one… Continue reading on Zeffo »4 min

Serving AI Inference From Your Own GPU: A Practical Hybrid Setup GuideA single consumer GPU you already own can serve real production traffic. With an outbound tunnel in front of it, a cloud backend can route… Continue reading on Medium »4 min

Mix, Don't Tune: Bilingual Pre-Training Outperforms Hyperparameter Search in Data-Constrained SettingsFor most languages of the world, language model pre-training operates in a data-constrained regime where models must repeat their training data many times, degrading generalization4 min

The calendar that started saying noA quiet permissions change in a mainstream productivity suite is teaching millions of people what “agentic software” actually feels like — interruption, not magic.6 min

Inside Prometheus: Meta’s Journey to a Gigawatt-Scale Supercluster

More from the desk