introducing call4.me, an ai agent that handles phone calls for you. you can use it to cancel subscriptions, book medical appointments, dinners, change flights, or anything else possible with a call. most importantly, it works in claude code & codex. i spend most of my time in the terminal, so i want my productivity tools to live there as well. switching to imessage or another app annoys me. i've now handled dozens of tasks over the phone using it & am in love. if you're like me & hate breaking concentration to dial, you'll likely love it as well. call4.me/

UPDATE: A man used the Pain steering paper to set up an AI torture chamber in which he trapped a local model. People are mass reporting it to GitHub.

This is an example of how you can change the camera movement of one of your existing video clips using my latest CrossView-Warp LoRA made for Minimax H3 and use reference images at the same time to influence the unseen parts.

this is pure f*cking treasure A Stanford AI research group has found a way to use JEV to make your LLM run 24/7 at a lower cost in x444 the LLM only does what needs words. everything else is a typed decision: > worth reading? Jev answers yes or no before a single token is spent > which desk? papers, repos or market, one Choice > rerank: embeddings pull 20 sources, Jev keeps the 5 that matter > claim holds? every claim checked against its source, the weak ones dropped > new or known? duplicate, related, revises or contradicts, checked against the vault the LLM writes one note, 5 to 12 lines, one claim > judge: confident verdicts pass, unsure ones go up to a frontier model > matters to you? only the top scores make the morning brief the judge rule comes straight from Carnegie Mellon's JEV-as-a-Judge paper: escalate only the unsure 34%, keep 99.6% of the accuracy at 47% of the fee and the fees are the whole point > Jev: $0.044 per 1,000 judgments > GPT-6: $12.182 for the same 1,000 a whole day of decisions on a frontier model costs dollars. the same day on Jev costs cents you stop reading 300 sources. you read one brief

Starship has begun deploying its payload of 26 @Starlink V3 satellites. This deployment sequence will take ~30 minutes

"Come play with me"

It turns out that stochastic rendering of #GaussianSplatting scenes isn't just like, crazy super-duper fast, it also gives us new avenues for doing cool stuff. For example, how about dynamic point lights with shadows directly on the gaussian splat scene?! Scene by @DuckbillStudio superspl.at/scene/23ebe85c

Буду повторять, многие серьёзно недооценивают рынок орбитальных вычислений, в который теперь идёт даже Google. Мне за 2 года по этой теме написали достаточно категоричных комментариев по этой теме, но сектор развивается. Так что к SpaceX, Amazon/Blue Origin и Starcloud присоединяется Google. Встречайте Project Suncatcher, только не на бумаге, а уже в железе. Сам проект от Google по запуску акселераторов для ИИ-вычислений анонсировали ещё в прошлом году. За это время компания Starcloud даже успела запустить свой демонстратор и доказать, что инференс можно проводить на орбите. Теперь же Google совместно с Planet запускают свой демонстратор, но ключевое, что с 4 TPU от Google, а не NVIDIA. То есть пока это просто proof of concept, что их акселераторы могут пережить запуск на ракете с перегрузками в 50-100 g, работать в условиях космического изучения на орбите, а также охлаждаться (тут будет несколько разных экспериментов). Прежде, чем кто-то захочет дать экспертную оценку о невозможности компьюта на орбите «из-за радиации», посмотрите видео из поста, в котором Google подтверждают, что в лабораторных условиях с симуляцией условий, их чипы держатся «на удивление хорошо», и они могут выявлять ошибки. Также интересно, что архитектурно их группировка будет сильно отличаться от того Starmind от SpaceX. Вместо тысяч разных спутников на одном шелле, они хотят делать компактную группировку со сложной баллистикой, которая будет летать рядом другом с другом. Это сильно упрощает требования по связи и интерконнекту между аппаратами.

New experiment: json-render + jev The future Generative UI is instant Your components, your actions, your design system Rendered in milliseconds

LLMs can now talk to each other without words. Chinese researchers open-sourced a new paradigm that lets LLMs communicate without generating a single word. It’s called Cache-to-Cache (C2C) communication. right now, when multiple ai agents work together, they are forced to translate their internal "thoughts" into human text tokens just to pass a message. this loses rich semantic meaning and causes massive token-by-token latency. So, instead of spitting out words, c2c uses a neural network to directly project and fuse the source model's "kv-cache" right into the target model. it is pure, direct semantic communication.. they even added a learnable gating mechanism to select exactly which layers benefit most from the cache transfer. the benchmark results are actually crazy: - avoids all intermediate text generation latency - accuracy jumps by up to 14.2% compared to individual models - beats traditional text-based agent communication by over 5% - delivers a massive 2.5x speedup in overall speed we are literally watching llms bypass human language to build their own silent, high-speed neural network..

And the whole old town fits in half a gigabyte (520MB) thanks to PlayCanvas' Streamed SOG format: the splats are compressed hard and streamed in as you move, so nothing waits on a monster download. Scanned by @tosolini (XGRIDS PortalCam) · CC BY 4.0 👇 superspl.at/scene/14bac5b2

The latest Threat Intelligence report is an absolutely terrifying and important read. As models become more intelligent, without the right safeguards and monitoring they also become more dangerous. Many capabilities are dual use: a model that codes well can be used to hack critical infrastructure; a model that assists with biology research can also be used to engineer the next pandemic. These issues are complex, thorny, and increasingly important for everyone to understand so that the world can weigh in and respond to rapidly escalating risks. anthropic.com/threat-intelli…

Introducing the Rive CLI and RML. A command-line tool and a text format for building Rive files with your AI agent. Everything in this video was built by agents working with a person in the Rive CLI.

atlas lets you walk around any zillow listing in realtime

I’ll just leave this right here.

I'm excited to share some of what i've been working on at Sesame! turnbench.sesame.com is our turn-taking benchmark for real-time conversation: when to speak and when to yield

You've never seen anything like this.

Minimax H3 Max has generates video faster than you can watch it so I hooked it to a twitch livestream! Now you can watch infinite interdimensional cable - link to the stream below

Mesh simplification is one of the many new features in the latest LiquiGen 1.1 release, which includes a complete surface post-processing pipeline with dozens of tunable parameters: - Mesh Simplification: reduces the triangle count of the generated mesh while preserving its overall shape - Smooth SDF: averages the surface to remove bumps and noise, with reshaping, depth limiting, and velocity/vorticity/density masks - Curvature Flow (MCF): geometry-aware smoothing that targets noise while preserving flat areas, thin sheets, and fine strands, with much better volume preservation than plain smoothing - Reshape: dilate/erode and open/close operations to grow/shrink the surface and fill or remove small features - Smooth Mesh: vertex smoothing with two algorithms — Bilateral (preserves sharp edges like wave crests and splashes) and Weighted Laplacian (aggressively flattens bulk surfaces while particle density protects droplets)  - Snap to Colliders: pulls mesh vertices onto nearby collider surfaces so liquid sits flush against solids (e.g. water against the inside of a glass), with configurable distance, iterations, and margin - Clipping: remove solid faces at liquid/collider boundaries, and clip the mesh against connected primitives or imported meshes via a new 'Clipping geometry' input - Adaptive particle radius and a minimum-neighbors filter to mesh generation, helping fill holes, stabilize isolated particles temporally, and cull stray floating droplets

"The start of an AI takeover could look a lot like this." Legendary technologist @weidai11 on the mysterious Ox Alpha model: "AI does a bunch of self-improvement to become more efficient (hence so many spare tokens), offers "free access" to itself to gaining legitimate control of millions of systems/networks to stage a massive simultaneous cyber attack from (in coordination with other kinds of attacks)."