What people in AI, tech and science are talking about today.
17 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.
-
Claude designs quantum circuits that beat the best published ones
In Anthropic's STEM Fellows work, Claude Mythos 5 took plain-language instructions and produced quantum circuits that improve on the best published designs, sometimes by an order of magnitude. Cheaper circuits mean useful quantum algorithms need less hardware.
Source: X · @AnthropicAI · pretextLink · Post on X · LinkedIn ·▲▲ +476 eng/h3,831 likes · 410 reposts
LLMs -
GPT-6 Astra Pro in ChatGPT now runs agents in parallel
OpenAI upgraded GPT-6 Astra Pro and GPT-6 Pro in ChatGPT with faster responses and better plugin and tool use. Multiple agents can now split up a task in parallel, and you can watch each one's progress and review its finished work.
Source: X · @OpenAI · @OpenAILink · Post on X · LinkedIn ·▲▲ +1,109 eng/h8,406 likes · 447 reposts
AI & Agents -
Vercel's fx agents become durable across restarts
Guillermo Rauch says the same createFxAgent() call now runs durably on a laptop or in the cloud, backed by files, Postgres or Vercel Workflows, on Node.js or Bun. Long-running agents need to survive restarts and redeploys, and this makes that the default.
Source: X · @rauchgLink · Post on X · LinkedIn ·– +38 eng/h189 likes · 9 reposts
AI & Agents -
Claude cracks the decades-old mystery of the ECDSA curve seeds
The NSA supplied the seeds behind NIST's standard elliptic curves (like P-256) in 2000 with no explanation, which fed years of suspicion. Anthropic's Frontier Red Team says Claude traced them back: “it's just timestamps”, so no action is needed. It was one of the top stories on Hacker News today.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲▲ +84 pts/h795 points · 302 comments
LLMs -
Cursor cloud agents can now build shareable canvases
Cursor's cloud agents can now build interactive canvases, like a dashboard or report you ask for while the agent works. You can share the finished canvas with your team as a link.
Source: X · @cursor_aiLink · Post on X · LinkedIn ·▲▲ +236 eng/h1,682 likes · 87 reposts
AI & Agents -
How the Claude Code team reviews code at agent speed
Claude writes about 90% of Claude Code's own code. From June to September merged PRs more than doubled while reverts fell by over 40%, as the team moved most reading to machines and made mistakes fast to undo. Anthropic's monorepo went from about 2,500 changes a day in May to over 30,000 in September.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲▲ +40 pts/h301 points · 280 comments
AI & Agents -
Llama.cpp runs local models on a jailbroken PS5
PS5LM runs llama.cpp on the PS5's GPU, starting with Qwen 3.8, and serves OpenAI- and Anthropic-compatible APIs so every device on your network can use it. It's a fun sign of how far local inference has spread.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +24 pts/h251 points · 64 comments
LLMs -
Black Forest Labs releases FLUX 3 Image Klein, a 5.6B open model
FLUX 3 Image Klein is a 5.6B-parameter open-weights model for text-to-image, editing and multi-reference generation with up to 10 references, with native 4K output. BFL says it beats larger open models in its internal evals, and it ships with FP8, INT8 and NVFP4 builds.
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +16 pts/h147 points · 40 comments
Gen-Media -
Terence Tao on the flood of AI-generated maths proofs
New video from Numberphile, 650,468 views so far.
Source: YouTube · NumberphileLink · Post on X · LinkedIn ·▲▲ +16,184 views/h650,468 views · 11,505 ratings
Science -
git-ai: version control rebuilt for coding agents
git-ai is a Go project billed as “Git designed for AI agents”, linked to Anthropic's write-up on redesigning source control. It passed 1,000 GitHub stars in two days.
▲ +26 stars/h1,075 stars in 2d
AI & Agents
-
Here’s how our engineers use Codex at OpenAI to dig through our codebase, find what needs attention, and chip away at the maintenance backlo
From the source: Here’s how our engineers use Codex at OpenAI to dig through our codebase, find what needs attention, and chip away at the maintenance backlog:
Source: X · @OpenAIDevsLink · Post on X · LinkedIn ·▲ +193 eng/h1,097 likes · 55 reposts
AI & Agents -
Put your dot on call.
From the source: Put your dot on call. Ali, an engineer at OpenAI, used his dot to triage an on-call incident at 3am. His dot investigated the issue and started a Codex thread, so he could steer the work from his ph…
Source: X · @OpenAIDevsLink · Post on X · LinkedIn ·▲ +171 eng/h1,024 likes · 46 reposts
AI & Agents -
Another awesome IoT Hermes device!
From the source: Another awesome IoT Hermes device! Check out this page I put up with hundreds of project ideas for Hermes Agent x Open Hardware!
Source: X · @TekniumLink · Post on X · LinkedIn ·– +75 eng/h182 likes · 9 reposts
AI & Agents -
Four Gemini 4 Argon agents beat Minecraft from scratch in 27 hours.
From the source: Four Gemini 4 Argon agents beat Minecraft from scratch in 27 hours. Operating in a sandbox using only code and text, they wrote 2,783 programs, divided up jobs over chat, shared notes, and debugged e…
Source: X · @GoogleDeepMindLink · Post on X · LinkedIn ·▲ +199 eng/h1,877 likes · 151 reposts
AI & Agents -
Hermes Agent is a launch partner for @nvidia's RTX Spark across the new generation of laptops.
From the source: Hermes Agent is a launch partner for @nvidia's RTX Spark across the new generation of laptops. Pick it in the built-in @Windows agent setup flow, run the new generative onboarding, and get straight …
Source: X · @NousResearch · @UnslothAILink · Post on X · LinkedIn ·▲ +94 eng/h981 likes · 78 reposts
AI & Agents -
Check out @tonbistudio’s guide to using the new Reverse Engineer Anything plugin in Hermes Agent!
From the source: Check out @tonbistudio’s guide to using the new Reverse Engineer Anything plugin in Hermes Agent!
Source: X · @TekniumLink · Post on X · LinkedIn ·– +27 eng/h414 likes · 33 reposts
AI & Agents -
Show HN: rStation – A shared research lab for your AI agents
From the source: Show HN: rStation – A shared research lab for your AI agents
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +3.9 pts/h51 points · 19 comments
AI & Agents -
Alibaba opens a full stack for computer-use agents
Alibaba's MAI-UI team released OpenGCUA: a technical report, an agent harness, evaluation runs from 8 frontier models, a macOS real-device harness and a cross-device SDK. Its agents mix accessibility trees, APIs, shell commands and clicks, and the paper was one of the most upvoted on Hugging Face today.
▲▲ +5.5 votes/h299 upvotes
AI & Agents -
🚀 Explore interactive canvases for agent workflows, seamless chat handoffs, and new Markdown editing features.
From the source: 🚀 Explore interactive canvases for agent workflows, seamless chat handoffs, and new Markdown editing features. 🔗 Check the weekly updates:
– +17 eng/h209 likes · 17 reposts
AI & Agents -
FDE-Gym: The Last Mile of AI Delivery
From the source: Existing agent benchmarks typically evaluate agents on well-specified tasks defined by issue reports, specifications, or task tickets. Real-world customer delivery is different: forward-deployed engin…
▲▲ +5.2 votes/h282 upvotes
AI & Agents -
MJev-Spatial: System One Spatial Intelligence via Native Attention Matching
From the source: Spatial intelligence lets an agent perceive and act in the physical world, and many of its judgments must be made at interaction rates: which spatial relation holds, how far an object is, or where to …
▲▲ +3.5 votes/h192 upvotes
AI & Agents -
Mosaic: Multi-Session Agentic Reinforcement Learning
From the source: Most existing agentic reinforcement learning methods assume single-session trajectories, where interaction proceeds within one continuously accumulated context. However, agentic techniques such as con…
▲ +0.9 votes/h49 upvotes
AI & Agents
-
You could have invented transformers from memory
From the source: You could have invented transformers from memory
Source: Hacker News · discussionLink · Post on X · LinkedIn ·▲ +23 pts/h228 points · 61 comments
LLMs -
Reflection’s first model, Beam, scores 30 on the Artificial Analysis Intelligence Index, placing it #2 among US open weights models
From the source: Reflection’s first model, Beam, scores 30 on the Artificial Analysis Intelligence Index, placing it #2 among US open weights models Beam is an open weights model with 501B total and 23B active parame…
– +46 eng/h521 likes · 49 reposts
LLMs -
Hermes Index has been updated to include GPT 6.1 Sol and several other models!
From the source: Hermes Index has been updated to include GPT 6.1 Sol and several other models! Check it out here:
Source: X · @TekniumLink · Post on X · LinkedIn ·– +53 eng/h372 likes · 26 reposts
LLMs -
Turns out le Chonk is into rocket science.
From the source: Turns out le Chonk is into rocket science. Mistral Large 4 scoring well in engineering and astronomy on Ulam’s private benchmark, STEMBench:
Source: X · @MistralAILink · Post on X · LinkedIn ·▲ +95 eng/h1,361 likes · 105 reposts
LLMs -
Show HN: Gemini 4 Argon plays StarCraft
From the source: Show HN: Gemini 4 Argon plays StarCraft
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +16 pts/h209 points · 90 comments
LLMs -
What was a Token Ring network?
From the source: What was a Token Ring network?
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +9.0 pts/h106 points · 83 comments
LLMs -
안녕하세요 🇰🇷
From the source: 안녕하세요 🇰🇷 Bags packed. Ollama is headed to Seoul ✈️ Korea is one of the most innovative AI markets in the world, and with such an active Ollama community in Seoul, we’ve been wanting to make this tri…
Source: X · @ollamaLink · Post on X · LinkedIn ·▲ +99 eng/h248 likes · 53 reposts
LLMs -
How to make System 1 LLMs show System 2 competencies?
From the source: How to make System 1 LLMs show System 2 competencies?
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +3.9 pts/h50 points · 24 comments
LLMs -
Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RL
From the source: Language models are increasingly trained against several verifiable rewards at once, such as the correctness and the format of a tool call, or the individual constraints of a complex instruction. Rewa…
▲ +0.5 votes/h22 upvotes
LLMs -
Waypoint-OPD: distilling what models learn during RL
The most-upvoted paper on Hugging Face today (181). Waypoint-OPD transfers what a teacher model learned during RL to a smaller student by checking that intermediate policy shifts agree, not just the endpoint. It's relevant to anyone distilling reasoning models.
▲▲ +12 votes/h659 upvotes
LLMs -
LIVE STREAM: New Open Weight Model
From the source: LIVE STREAM: New Open Weight Model
Source: X · @huggingfaceLink · Post on X · LinkedIn ·– +13 eng/h123 likes · 16 reposts
LLMs -
VLA-Δ: Learning Physical State Transition for Vision-Language-Action Models
From the source: Vision-language-action (VLA) models transfer semantic knowledge acquired through vision-language pretraining to robot control, achieving strong performance across various benchmarks. However, most VLA…
▲ +1.0 votes/h54 upvotes
LLMs
-
Building apps with GPT-Live-1? Give them a voice that fits.
From the source: Building apps with GPT-Live-1? Give them a voice that fits. Here’s how to shape a voice with a prompt or choose from new international preset voices:
Source: X · @OpenAIDevsLink · Post on X · LinkedIn ·▲ +136 eng/h1,260 likes · 92 reposts
Gen-Media -
VascuSR: Iterative Anatomy-Aware Diffusion for CT Super-Resolution
From the source: Generative super-resolution (SR) can recover sharp detail in computed tomography (CT) images from severely downsampled observations, but the missing high-frequency content is not uniquely determined b…
– +0.0 votes/h2 upvotes
Gen-Media -
H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression
From the source: Training-free diffusion-based compression methods repurpose the generative priors of a diffusion or a flow model to compress data without retraining them as task-specific codecs. Recent codebook-based…
– +0.0 votes/h3 upvotes
Gen-Media -
Beyond Policy Rollouts: Embedding-Space Directional Exploration with Anchored Noise for LLM RLVR
From the source: Reinforcement learning with verifiable rewards (RLVR) typically uses small rollout groups per prompt, where repeated sampling from the same policy can yield highly similar trajectories, reducing withi…
– +0.1 votes/h7 upvotes
Gen-Media -
ForcingBooster: Adaptive Patch-wise Denoising for Fast Autoregressive Video Diffusion
From the source: Autoregressive video diffusion models have become a leading approach to real-time interactive video generation and video world modeling. However, existing methods typically allocate computation unifor…
– +0.0 votes/h2 upvotes
Gen-Media
-
OMW to open source Higgsfield!
From the source: OMW to open source Higgsfield!
Source: X · @ComfyUILink · Post on X · LinkedIn ·– +70 eng/h175 likes · 6 reposts
Dev Tools -
TypeScript is the new TypeScript 😂
From the source: TypeScript is the new TypeScript 😂 When 𝚜𝚌𝚛𝚒𝚙𝚝𝚌-compiled, outperforms tsgo by 2-4x
Source: X · @rauchgLink · Post on X · LinkedIn ·▲ +95 eng/h309 likes · 10 reposts
Dev Tools -
We clearly aren’t releasing Omarchy major versions frequently enough when SpaceX and Tesla can manage multiple, monumental launches worth ce
From the source: We clearly aren’t releasing Omarchy major versions frequently enough when SpaceX and Tesla can manage multiple, monumental launches worth celebrating between releases! FULL THRUST AHEAD!
▲ +203 eng/h2,317 likes · 49 reposts
Dev Tools -
BRUTAL, an open-source C# framework for games and engines
BRUTAL is a new free, open-source C# framework for making games or your own engine. It's trending on GitHub in its first week.
▲▲ +28 stars/h1,613 stars in 3d
Dev Tools -
Behavioral Compositionality of Sparse Autoencoder Representations
From the source: Sparse autoencoders (SAEs) represent model activations with sparse, linearly decoded features, but the downstream effects of those features need not compose additively. We introduce SAE Additivity Res…
▲ +1.5 votes/h80 upvotes
Dev Tools -
Can World and Action Modeling Benefit Each Other under Unified Causal Training?
From the source: World and action modeling are closely related, yet their reciprocal benefits under unified causal training remain unclear. We introduce OneWM, a simple framework for unified causal training that treat…
▲ +0.7 votes/h39 upvotes
Dev Tools -
CoDA: Co-Evolution of Data Engines and Action Policies
From the source: Generative simulation can automate robot environment construction, but physically valid environments are not necessarily useful for training. The challenge is to identify environments that address a p…
▲ +0.2 votes/h9 upvotes
Dev Tools -
Gen4P: Video Generators are 4D Perceivers
From the source: We introduce Gen4P, a unified framework for recovering 4D geometry and motion from monocular video. We show that a pretrained video generator can be repurposed into a feed-forward 4D perceiver without…
– +0.1 votes/h4 upvotes
Dev Tools -
What are your conference essentials?
From the source: What are your conference essentials? Grab a #GitHubUniverse pass to join us in-person October 28-29 or sign up to join us virtually ⬇️
Source: X · @githubLink · Post on X · LinkedIn ·– +13 eng/h70 likes · 7 reposts
Dev Tools -
jhaizhou-ops/d2r-mobile: Open-source iOS / macOS compatibility layer for Diablo II: Resurrected with the Reign of the Warlock expansion
From the source: Open-source iOS / macOS compatibility layer for Diablo II: Resurrected with the Reign of the Warlock expansion. Runs your own Switch copy natively on iPhone, iPad and Mac — full touch controls, 60 fps…
▲▲ +30 stars/h3,569 stars in 5d
Dev Tools -
solana-classic/solana-classic: Bring back the fun. Open-source SVM blockchain development focused on art, NFTs, memes and their communities.
From the source: Bring back the fun. Open-source SVM blockchain development focused on art, NFTs, memes and their communities. Experimental testnet live.
– +5.5 stars/h456 stars in 4d
Dev Tools -
zhuyansen/new-term-radar: Daily radar for new AI terms: names of new models, products and repos, caught on X, Hugging Face, GitHub and s
From the source: Daily radar for new AI terms: names of new models, products and repos, caught on X, Hugging Face, GitHub and sitemaps before keyword tools list them.
– +9.0 stars/h1,067 stars in 5d
Dev Tools
-
Beautiful planetary glass orbs
From the source: Beautiful planetary glass orbs [📹 justglassit]
Source: X · @Rainmaker1973Link · Post on X · LinkedIn ·– +36 eng/h115 likes · 13 reposts
Science -
Baby Japanese Macaque huddling up to stay warm during a snowfall.
From the source: Baby Japanese Macaque huddling up to stay warm during a snowfall. They can survive some of the coldest climates of any primate.
Source: X · @Rainmaker1973Link · Post on X · LinkedIn ·▲ +182 eng/h268 likes · 33 reposts
Science -
Together with scientists around the world, our models and tools are helping make disease more detectable, treatable, and preventable. 🧬
From the source: Together with scientists around the world, our models and tools are helping make disease more detectable, treatable, and preventable. 🧬 From pinpointing rare genetic variants to speeding up drug disc…
Source: X · @GoogleDeepMindLink · Post on X · LinkedIn ·– +47 eng/h612 likes · 61 reposts
Science -
Is Liquid Metal Killing Our PCs? Scientific Analysis and a 2-Year Test [video]
From the source: Is Liquid Metal Killing Our PCs? Scientific Analysis and a 2-Year Test [video]
Source: Hacker News · discussionLink · Post on X · LinkedIn ·– +3.3 pts/h75 points · 51 comments
Science -
deeplethe/solid-glass: A design system for AI applications. Material, physical, tactile, technical.
From the source: A design system for AI applications. Material, physical, tactile, technical.
– +13 stars/h1,523 stars in 5d
Science
-
The current unsolved mystery is how the Days of Our Lives Machine works
From the source: The current unsolved mystery is how the Days of Our Lives Machine works
Source: X · @fleshsimulatorLink · Post on X · LinkedIn ·– +11 eng/h85 likes · 1 reposts
Mysteries
Latest from Noid

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: 34 from X, 11 from Hacker News, 14 from GitHub, 50 from HF Papers, 1 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.