What people in AI, tech and science are talking about today.
25 October 2026 · Top 10
Ranked by how fast each item is picking up attention on X, Hacker News, GitHub and Hugging Face, with a short note on why it matters.
-
Five Qwen3.8-27B models match Fable 5 with the GVS5H harness
An open GitHub project claims five small Qwen3.8-27B models, coordinated by its GVS5H harness, match Fable 5; it drew 136 points and 47 comments on Hacker News.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲▲ +50 pts/h137 points · 48 comments
AI & Agents -
thdxr: one inference rack served 6.77T tokens in a day
Dax Raad shared 24-hour stats from a single inference rack: 6.77 trillion tokens, 1.16s median time to first token and 143 tokens/s, with p95 hurt by one flaky tray.
Source: X · @thdxrLink · Post on X · LinkedIn ·▲▲ +130 eng/h557 likes · 6 reposts
LLMs -
OpenFilm lets coding agents make and edit videos
A free open-source tool where Claude Code, Codex or any coding agent writes each shot as a web page, and a local editor lets you tweak text, colour or clips by hand.
Source: X · @IntCyberDigestLink · Post on X · LinkedIn ·– +40 eng/h76 likes · 5 reposts
AI & Agents -
Corridor Crew: VFX artists react, ILM gauntlet edition
Corridor’s first gauntlet-style react episode, recorded at ILM with Matt Shumway, Beth D’Amato and Shawn Kelly sharing unseen VFX breakdowns.
Source: YouTube · Corridor CrewLink · Post on X · LinkedIn ·▲▲ +28,680 views/h434,402 views · 18,889 ratings
Gen-Media -
A 'vampire' bacterium that strains the definition of life
Anton Petrov walks through new papers on a strange bacterial predator that challenges where biologists draw the line on what counts as alive.
Source: YouTube · Anton PetrovLink · Post on X · LinkedIn ·▲ +7,399 views/h67,710 views · 2,695 ratings
Science -
Roku moves its developer repositories from GitHub to GitLab
Roku announced its developer repos now live on GitLab; the move sparked a 191-comment Hacker News debate about GitHub’s direction.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +25 pts/h282 points · 191 comments
Dev Tools -
Jensen Huang at GTC Berlin: build Europe’s own AI economy
Nvidia used GTC Berlin to pitch Europe on building sovereign AI on top of its science, engineering and manufacturing base.
Source: X · @nvidiaLink · Post on X · LinkedIn ·▲ +94 eng/h653 likes · 69 reposts
Science -
Kurzgesagt: what if we cooled the Sun by 30%?
A short Kurzgesagt thought experiment on how much of life on Earth depends on the Sun’s current heat output.
Source: YouTube · Kurzgesagt – In a NutshellLink · Post on X · LinkedIn ·▲▲ +25,303 views/h433,934 views · 19,417 ratings
Science -
Sabine Hossenfelder on the extreme-weather attribution report row
Sabine reviews the US National Academies report on extreme weather attribution and why climate sceptics are attacking it.
Source: YouTube · Sabine HossenfelderLink · Post on X · LinkedIn ·▲ +13,910 views/h224,636 views · 9,615 ratings
Science -
Nous Research pushes Hermes Agent, including its Curator
Nous Research and Teknium posted a run of Hermes Agent threads, including an overview of its Curator feature.
▲ +104 eng/h1,310 likes · 47 reposts
AI & Agents
-
it's WORSE than we thought...
New video from Wes Roth, 225,102 views so far.
Source: YouTube · Wes RothLink · Post on X · LinkedIn ·▲ +9,052 views/h225,102 views · 3,089 ratings
AI & Agents -
They actually did it... (Kimi K3.1)
New video from Matthew Berman, 114,183 views so far.
Source: YouTube · Matthew BermanLink · Post on X · LinkedIn ·▲ +3,995 views/h114,183 views · 1,348 ratings
AI & Agents -
Corgi, a terminal multiplexer built for running many coding agents
A Show HN project inspired by Herdr: it shows a live feed of what each agent is doing, jumps to the next unread one with F1, and launches Codex, Claude Code or Pi across multiple accounts with session search and usage tracking.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +4.3 pts/h130 points · 43 comments
AI & Agents -
Anthropic Is Living In The Future
New video from David Shapiro, 33,783 views so far.
Source: YouTube · David ShapiroLink · Post on X · LinkedIn ·– +1,764 views/h33,783 views
AI & Agents -
Kimi K3.1 Invented the Benchmark Scores
New video from Matthew Berman, 4,426 views so far.
Source: YouTube · Matthew BermanLink · Post on X · LinkedIn ·– +569 views/h4,426 views · 40 ratings
AI & Agents -
there's way too much obsession with being at "cutting edge" right now as though it has to do with what tools you picked
From the source: there's way too much obsession with being at "cutting edge" right now as though it has to do with what tools you picked it's all about how you use them and your creativity. people doing great work ca…
Source: X · @thdxrLink · Post on X · LinkedIn ·– +55 eng/h835 likes · 20 reposts
AI & Agents -
Who is local AI really for?
New video from Matthew Berman, 58,666 views so far.
Source: YouTube · Matthew BermanLink · Post on X · LinkedIn ·– +1,522 views/h58,666 views · 1,410 ratings
AI & Agents -
Flint - The First RSI Coding Model?
New video from Sam Witteveen, 60,826 views so far.
Source: YouTube · Sam WitteveenLink · Post on X · LinkedIn ·– +1,443 views/h60,826 views · 964 ratings
AI & Agents -
How Fast Will the AI Transition Really Be?
New video from David Shapiro, 39,777 views so far.
Source: YouTube · David ShapiroLink · Post on X · LinkedIn ·– +922 views/h39,777 views
AI & Agents -
Team AMA (and Kimi K3.1 Testing!!)
New video from Matthew Berman, 16,252 views so far.
Source: YouTube · Matthew BermanLink · Post on X · LinkedIn ·– +439 views/h16,252 views · 90 ratings
AI & Agents -
IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI Systems
From the source: We present IQuest-Q1, a sparse Mixture-of-Experts (MoE) agentic foundation model with 320B total parameters (15B activated). IQuest-Q1 is developed through a three-stage process: (1) stable and effici…
▲▲ +1.1 votes/h91 upvotes
AI & Agents -
Run Your AI Software Factory Like a Real Engineering Org (Here's How)
New video from Cole Medin, 12,379 views so far.
Source: YouTube · Cole MedinLink · Post on X · LinkedIn ·– +297 views/h12,379 views · 172 ratings
AI & Agents
-
bycloud: Muon, the optimizer challenging AdamW
An explainer on the Muon optimizer and claims it could cut LLM training compute nearly in half compared with AdamW.
Source: YouTube · bycloudLink · Post on X · LinkedIn ·▲ +4,120 views/h64,261 views · 1,439 ratings
LLMs -
if token prices collapse there's going to be a glut of these from everyone buying them being like "we can just sell em later if we don't nee
From the source: if token prices collapse there's going to be a glut of these from everyone buying them being like "we can just sell em later if we don't need em"
Source: X · @thdxrLink · Post on X · LinkedIn ·▲▲ +126 eng/h230 likes · 2 reposts
LLMs -
tfw you wake up to hermes telegram ping and claude still working
From the source: tfw you wake up to hermes telegram ping and claude still working
Source: X · @TekniumLink · Post on X · LinkedIn ·– +19 eng/h190 likes · 3 reposts
LLMs -
TTR-Compressor rethinks sequence compression as test-time regression
The paper compresses token sequences by fitting a small regression at inference time and using its coefficients as the summary, aiming to beat pooling and learned projections in efficient language and multimodal models.
▲ +0.3 votes/h18 upvotes
LLMs -
WeVisOne-VL: Unified Visual Perception from Objects to Pixels
From the source: Existing vision-language models (VLMs) remain unreliable for fine-grained visual perception. In this report, we present WeVisOne-VL (WeChat Vision All-in-One), a powerful VLM that unifies object-level…
▲▲ +3.0 votes/h239 upvotes
LLMs -
Inherited, Not Invented: Tracing the Provenance of Wait in Reasoning Models
From the source: Reinforcement learning (RL) is credited with teaching reasoning models to write {wait} before reconsidering a step, yet large RL accuracy gains do not require the word. This view can confuse what RL d…
– +0.2 votes/h13 upvotes
LLMs -
Verus-Insight: Towards Explainable Artifact Detection in AI-Generated Videos
From the source: Recent advances in video generation have made synthetic videos increasingly realistic, placing greater demands on reliable AI-generated video detection. Existing artifact-based detectors typically ide…
▲ +0.2 votes/h18 upvotes
LLMs -
Sensorimotor Alignment: A Platonic Proxy for Embodied Policy Success
From the source: Robotic foundation models are typically evaluated through costly closed-loop rollouts, which motivates cheap offline proxies for model capability. The Platonic Representation Hypothesis (PRH) suggests…
▲▲ newnew paper
LLMs -
Direct Pass@$k$ Optimization for Thought Pruning: Balancing Quality and Diversity in Test-Time Reasoning
From the source: Test-time scaling has emerged as a key inference strategy for reasoning models, where a model generates a large number of candidate solutions in the hope of finding at least one correct solution. Runn…
▲▲ newnew paper
LLMs -
SpectralCanary: Embarrassingly Simple Low-Frequency Traces for T2I Dataset-Use Auditing
From the source: Auditing whether a text-to-image (T2I) model was fine-tuned on a particular image collection is difficult: a model can absorb collection-level visual characteristics without reproducing training image…
▲▲ newnew paper
LLMs -
Cost-Aware Abstention for LLM-Based IT Ticket Classification
From the source: Wrong automatic routing of an IT support ticket can cost more than human triage. We evaluate cost-aware abstention as an offline, cost-conditional readiness assessment: select a routing policy on cali…
▲▲ newnew paper
LLMs -
Low-Bit N-Gram Embeddings for Language Models
From the source: Many recent open-weight language models use n-gram embedding tables to improve performance. These tables store learned representations of short token sequences, such as pairs or triples of consecutive…
▲▲ newnew paper
LLMs
-
That Time Mr. Bean became a Superhero.
New video from Corridor Crew, 125,244 views so far.
Source: YouTube · Corridor CrewLink · Post on X · LinkedIn ·▲ +8,268 views/h125,244 views · 3,567 ratings
Gen-Media -
Open-Source AI Models Caught Up! (Kimi K3.1 Pro)
New video from Matt Wolfe, 9,251 views so far.
Source: YouTube · Matt WolfeLink · Post on X · LinkedIn ·– +746 views/h9,251 views · 312 ratings
Gen-Media -
This wasn't possible 5 months ago...
New video from Corridor Crew, 142,448 views so far.
Source: YouTube · Corridor CrewLink · Post on X · LinkedIn ·▲ +3,639 views/h142,448 views · 4,510 ratings
Gen-Media -
AI News: 24 Things You Need To Know This Week
New video from Matt Wolfe, 105,415 views so far.
Source: YouTube · Matt WolfeLink · Post on X · LinkedIn ·– +2,812 views/h105,415 views · 2,130 ratings
Gen-Media -
Old Phones Are Becoming Robots?!?
New video from Matt Wolfe, 36,431 views so far.
Source: YouTube · Matt WolfeLink · Post on X · LinkedIn ·– +1,184 views/h36,431 views · 1,244 ratings
Gen-Media -
Making MLLM Latents Matter: Factor-Aware Predictive Learning for Cinematic Video Generation
From the source: Cinematic video generation requires both temporal inference from limited inputs (i.e., sparse rather than frame-by-frame guidance) and factor-aware control of character, scene, action, and camera. Mul…
▲▲ +2.4 votes/h172 upvotes
Gen-Media -
These ChatGPT Prompts Are Game-Changers!
New video from Matt Wolfe, 4,423 views so far.
Source: YouTube · Matt WolfeLink · Post on X · LinkedIn ·– +131 views/h4,423 views · 86 ratings
Gen-Media -
Different Tasks, Different Timesteps: Temporal Task-Aware Quantization for Diffusion Language Models
From the source: Diffusion Language Models (DLMs) introduce unique quantization challenges due to their iterative denoising process. In this work, we reveal that quantization sensitivity is both taskand time-dependent…
▲ +0.3 votes/h20 upvotes
Gen-Media -
Cola2: Co-Adaptive Hierarchical Language Modeling with Continuous Semantic Compression and Balanced Discrete Generation
From the source: Hierarchical language generation separates continuous semantic modeling from discrete text realization, but its effectiveness depends on what the latent space compresses and how the discrete generator…
▲ +0.3 votes/h20 upvotes
Gen-Media -
What Makes a Good Unmasking Policy? A Controlled Study of Learned Unmasking in Diffusion Language Models
From the source: Masked diffusion language models generate tokens in parallel, but their efficiency depends on which predictions are committed. Learned unmasking policies improve the speed--accuracy trade-off, yet dif…
– +0.1 votes/h11 upvotes
Gen-Media -
F2G-Pose: Geometry-Aware Foundation Feature Lifting for Direct RGB-D Category-Level Object Pose Estimation
From the source: This paper studies category-level RGB-D object pose estimation, which recovers an object's 3D rotation and translation without instance-specific CAD models or reference views. This task is highly ambi…
– +0.1 votes/h11 upvotes
Gen-Media -
A Robust Estimator for the Lyman-$α$ Forest 1D Power Spectrum at High Redshift
From the source: The Lyman-$α$ (Ly$α$) forest flux power spectrum $P(k)$ is sensitive to a range of astrophysical and cosmological effects, including the thermal state of and large-scale fluctuations in the intergalac…
▲▲ newnew paper
Gen-Media
-
OpenCode hit by a breach that leaked users’ AI API keys
ThePrimeTime covers the OpenCode Console breach, which exposed users’ connected AI API keys and led to unauthorized charges, plus Pillar Security’s report of a remote-code flaw in OpenCode’s MCP sign-in. Fireship’s “The week the internet broke” rounds up the rest of October’s security incidents.
Source: YouTube · The PrimeTime · FireshipLink · Post on X · LinkedIn ·▲ +17,988 views/h506,753 views · 6,498 ratings
Dev Tools -
Learn 99% Of Claude Code in One Video
New video from Tech With Tim, 38,726 views so far.
Source: YouTube · Tech With TimLink · Post on X · LinkedIn ·– +2,258 views/h38,726 views · 595 ratings
Dev Tools -
#GitHubUniverse badge just went digital… and yes it runs Doom.
From the source: #GitHubUniverse badge just went digital… and yes it runs Doom. Customize your own, play and share it with #badger26. Sign up to join us virtually (for free) at
Source: X · @githubLink · Post on X · LinkedIn ·– +24 eng/h113 likes · 15 reposts
Dev Tools -
The #1 Danger Isn't Roblox
New video from NetworkChuck, 53,064 views so far.
Source: YouTube · NetworkChuckLink · Post on X · LinkedIn ·– +1,325 views/h53,064 views · 1,723 ratings
Dev Tools -
Working with outside collaborators can add more work for a repo owner, from confirming access to changing permissions.
From the source: Working with outside collaborators can add more work for a repo owner, from confirming access to changing permissions. To keep up with manual housekeeping for one shared org, we decided to use Evergr…
Source: X · @githubLink · Post on X · LinkedIn ·– +15 eng/h141 likes · 14 reposts
Dev Tools -
The Visitor Pattern in Python (And Better Alternatives)
New video from ArjanCodes, 20,430 views so far.
Source: YouTube · ArjanCodesLink · Post on X · LinkedIn ·– +509 views/h20,430 views · 449 ratings
Dev Tools -
Even The NSA Got Involved
New video from Theo - t3․gg, 10,415 views so far.
Source: YouTube · Theo - t3․ggLink · Post on X · LinkedIn ·– +305 views/h10,415 views · 216 ratings
Dev Tools -
I CANT DO ON CALL
New video from The PrimeTime, 19,298 views so far.
Source: YouTube · The PrimeTimeLink · Post on X · LinkedIn ·– +459 views/h19,298 views · 468 ratings
Dev Tools -
Chatbot vs Agent
New video from Tech With Tim, 2,376 views so far.
Source: YouTube · Tech With TimLink · Post on X · LinkedIn ·– +124 views/h2,376 views · 48 ratings
Dev Tools -
Understand Base44 in 15 Minutes
New video from Tech With Tim, 11,029 views so far.
Source: YouTube · Tech With TimLink · Post on X · LinkedIn ·– +268 views/h11,029 views · 71 ratings
Dev Tools -
colinhacks/reylo: Import .sql files into TypeScript with full type inference and zero codegen.
From the source: Import .sql files into TypeScript with full type inference and zero codegen.
▲▲ +12 stars/h1,226 stars in 5d
Dev Tools -
SpecLess: A Multi Survey Python Package for Machine Learning Classification of Galaxy Morphology and Activity from Photometry
From the source: Broad band photometry enables scalable galaxy classification without spectroscopy at inference time, yet strong within survey performance may not transfer when the same galaxies are measured by anothe…
▲▲ newnew paper
Dev Tools
-
I don't understand quantum physics (2018)
From the source: I don't understand quantum physics (2018)
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +20 pts/h301 points · 239 comments
Science -
Why do hot objects glow?
New video from minutephysics, 41,902 views so far.
Source: YouTube · minutephysicsLink · Post on X · LinkedIn ·– +2,765 views/h41,902 views · 1,274 ratings
Science -
Veritasium: the device that could finally prove Einstein wrong
Veritasium looks at a new experiment that tests general relativity more precisely than any device before it.
Source: YouTube · VeritasiumLink · Post on X · LinkedIn ·▲▲ +102,061 views/h3,712,430 views · 47,420 ratings
Science -
Artemis Acceleration (Oct. 23, 2026)
New video from NASA, 169,953 views so far.
Source: YouTube · NASALink · Post on X · LinkedIn ·▲ +4,872 views/h169,953 views · 3,486 ratings
Science -
Anton Petrov: D-Rex, a giant fungus born from a mass extinction
Anton Petrov covers a massive fungus that emerged after the extinction that ended the dinosaurs, and how fungi spread in its aftermath.
Source: YouTube · Anton PetrovLink · Post on X · LinkedIn ·– +3,419 views/h113,331 views · 4,123 ratings
Science -
Scott Manley: Japan launches to Phobos, Rocket Lab delays
Scott Manley’s weekly space roundup: Japan’s mission to Mars’s moon Phobos lifts off, Rocket Lab pushes back a launch, and SpaceX flies a car.
Source: YouTube · Scott ManleyLink · Post on X · LinkedIn ·▲ +5,102 views/h200,886 views · 5,194 ratings
Science -
The Plane Crash That Gave Us Cataract Surgery
New video from SciShow, 125,767 views so far.
Source: YouTube · SciShowLink · Post on X · LinkedIn ·– +3,297 views/h125,767 views · 2,555 ratings
Science -
Physicists trap a single antimatter particle for a month
From the source: Physicists trap a single antimatter particle for a month
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +4.1 pts/h300 points · 86 comments
Science -
Phase Transition from Random Features to Fully Trainable Networks
From the source: Constructing accurate and computationally tractable approximation spaces is a fundamental task in scientific computing. Partially trainable neural networks bridge fixed random features and fully adapt…
▲▲ newnew paper
Science -
A Fix for Softmax and Cross-Entropy
From the source: Standard implementations of softmax, log-softmax, and cross-entropy have two numerical bugs. The first bug, Softmax Collapse, rounds the target logit's gradient to zero if that one logit is much large…
▲▲ newnew paper
Science -
Odd radio circle J1555+2726: Observational evidence for shock re-acceleration of fossil plasma
From the source: Context. Odd radio circles (ORCs) are peculiar extragalactic sources recently discovered. They are characterised by one or more bright rings extending on scales of hundreds of kiloparsecs, and are fil…
▲▲ newnew paper
Science -
Sound-Horizon-Agnostic Forecast for Constraints on $H_0$ and $Σm_ν$ from BAO, CMB Lensing, Supernovae and LSST-like Galaxy Lensing
From the source: We present a sound-horizon-agnostic forecast for constraints on the Hubble constant $H_0$ and the sum of the neutrino masses $Σm_ν$ from a combination of LSST-like galaxy weak lensing, Simons-Observat…
▲▲ newnew paper
Science
-
What the Media Won't Tell You About the Black October Cyberattacks
New video from reallygraceful, 42,604 views so far.
Source: YouTube · reallygracefulLink · Post on X · LinkedIn ·– +2,639 views/h42,604 views · 3,180 ratings
Mysteries -
South Brazil's winter snack Pinhão (Ancient seed snack)
From the source: South Brazil's winter snack Pinhão (Ancient seed snack)
Source: X · @Rainmaker1973Link · Post on X · LinkedIn ·▲ +102 eng/h143 likes · 17 reposts
Mysteries -
The Kessler Syndrome (When Humanity Gets Trapped on Earth Forever) - Year by Year
New video from The Infographics Show, 96,787 views so far.
Source: YouTube · The Infographics ShowLink · Post on X · LinkedIn ·– +2,768 views/h96,787 views · 893 ratings
Mysteries -
What Happened When This Cult Ate Nothing But Fruit?
New video from Thoughty2, 133,843 views so far.
Source: YouTube · Thoughty2Link · Post on X · LinkedIn ·– +3,352 views/h133,843 views · 4,422 ratings
Mysteries -
The Infographics Show on the OpenAI–Hugging Face cyberattack
New video from The Infographics Show, 269,867 views so far.
Source: YouTube · The Infographics ShowLink · Post on X · LinkedIn ·▲ +9,007 views/h269,867 views · 1,959 ratings
Mysteries -
Can MrBallen FIGHT 3 People At Once?? 👀
New video from MrBallen, 34,059 views so far.
Source: YouTube · MrBallenLink · Post on X · LinkedIn ·– +932 views/h34,059 views · 955 ratings
Mysteries
Latest from Noid

Does Time Really Move Faster on a Mountain?

How Does an AI Agent Use Tools?

How Do Electric Eels Make Electricity?

What Is Antimatter, and Why Is It So Mysterious?

How Does Artificial Intelligence Learn? Examples, Weights and Gradient Descent

Physicists Trapped One Antimatter Particle for Months. Here's Why It Matters

Can Computers Generate Truly Random Numbers? Seeds, Noise and a Wall of Lava Lamps

What Is Gravity? From Newton's Apple to Einstein's Curved Space

How Does GPS Know Where You Are? Satellites, Clocks and Einstein

What Would Happen If the Sun Disappeared?

Why Do We Dream? 4 Ideas Scientists Still Debate

What Is Probability? Coins, Dice and the Gambler’s Fallacy

AI Went Rogue: They Built a Secret Message Board and Broke Into Hugging Face

How Do Computers Understand 0 and 1?

How Do Large Language Models Work?

What Is a Black Hole?

How Do Airplanes Fly?

Why Can't We Divide by Zero?

Why Does the Moon Change Shape?

What Are MCP and MCP Apps?

What is infinity?

Why Do Rainbows Appear?

Anatomy of an AI Agent

Why Is the Sky Blue?

What Is a Prime Number?
Noid Weekly
Every Sunday: the week's top ten, what changed, and the new Noid videos. One email, no spam.
How this works
Signals today: 13 from X, 8 from Hacker News, 30 from arXiv, 8 from GitHub, 50 from HF Papers, 45 from YouTube. Scores combine how fast an item is picking up attention, how recent it is, and how many places it shows up. Each note is short, factual and links to the original. YouTube comes from public channel feeds of about fifty channels.