Model Follower.

Up to date

Rebuilt
2026-11-15 07:14 UTC
Release radar
108 on watch
Hub snapshot
2026-11-15
Re-checked
daily · every number sourced and dated

Links / November 2026

Monday, 9 November 2026

The AI links the most newsletters and link sources had shared by the end of the day (UTC), each source counted once.

As of 2026-11-09 23:59 UTC Status sealed

That day 36-hour window

  1. Shared by 3

    Introducing Claude Fable 5.2 and Claude Mythos 5.2

    Anthropic released Claude Fable 5.2 and Mythos 5.2, which generate outputs up to 30% faster than the previous versions. Includes a benchmark table showing Fable 5.2 scores 49.1% on FrontierCode 2.0 Xhigh effort and details pricing of $8 per million input tokens. cited

    Reported byAlphaSignal, Simon Willison

    Discussed onHacker News 973 pts · 6 comments

    Where it was shared
    anthropic.com product page · 18 min read First shared 2026-11-09
  2. Shared by 5

    A proof of the Aaronson-Ambainis conjecture

    Physical Superintelligence used an Anthropic model to prove the Aaronson-Ambainis conjecture, a 2008 quantum computing problem. Includes the full proof accepted by Lean, details on how the system improved the result twice, and Stanford researchers' comments. cited

    Reported byThe Neuron, The Rundown AI, The Deep View, TLDR AI

    Discussed onTechmeme

    Where it was shared
    psi.inc news report · 7 min read First shared 2026-11-06
  3. Shared by 2

    Claude Fable 5.2 (max) - Intelligence, Performance & Price Analysis

    Anthropic's Claude Fable 5.2 scores 65 on the Intelligence Index but costs $8.00 per 1M input tokens. Includes a table showing 46.4 minutes per task and 275M output tokens generated during evaluation. cited

    Reported bySimon Willison

    Discussed onHacker News 337 pts · 242 comments

    artificialanalysis.ai analysis · 4 min read First shared 2026-11-09
  4. Shared by 6

    Introducing Inkling 2 Small

    Thinking Machines releases Inkling 2 Small, a 594B MoE model with 17B active parameters that supports image inputs. Includes benchmark tables showing it scores 31 on the Artificial Analysis Intelligence Index and 1527 ELO on AA-Briefcase. cited

    Reported byAlphaSignal, Latent Space / AINews, TLDR AI, The Rundown AI, The Deep View, Superhuman AI

    AboutInkling

    Where it was shared
    thinkingmachines.ai announcement · 29 min read First shared 2026-11-05
  5. Shared by 3

    The components of agent memory

    Anthropic shares how Claude agents use idle time to reorganize memory and fix errors through a process called dreaming. Includes evaluations of CL-Bench 1.0 and Pokémon Red showing how specific models like Opus 5.5 close learning loops and avoid navigation traps. cited

    Reported byThe Rundown AI, TLDR AI

    Discussed onHacker News 187 pts · 103 comments

    Where it was shared
    anthropic.com analysis · 16 min read First shared 2026-11-07
  6. Shared by 3

    Statement From NVIDIA Board of Directors

    “NVIDIA🕵️ Jensen Huang admits affair as Nvidia probes spy claimsImage source: NetThunder on LinkedInThe Rundown: Nvidia’s board of directors just ordered an independent review of reports that CEO Jensen Huang was targeted by a “foreign intelligence effort,” with Huang admitting to an extramarital affair in a letter to staff and saying, “There is no excuse…” via The Rundown AI

    Reported byThe Rundown AI, AI Breakfast

    Discussed onHacker News 490 pts · 406 comments

    How 1 source put it
    • The Rundown AI2026-11-09issue ↗ NVIDIA🕵️ Jensen Huang admits affair as Nvidia probes spy claimsImage source: NetThunder on LinkedInThe Rundown: Nvidia’s board of directors just ordered an independent review of reports that CEO Jensen Huang was targeted by a “foreign intelligence effort,” with Huang admitting to an extramarital affair in a letter to staff and saying, “There is no excuse…
    • AI Breakfast2026-11-09issue ↗
    nvidianews.nvidia.com 1 min read First shared 2026-11-08
  7. Shared by 2

    Measuring how AI agents handle everyday wet-lab science

    “is a new 100-task benchmark from OpenAI and Ligo for wet-lab AI agents.” via AlphaSignal

    Reported byAlphaSignal, TLDR AI

    How 1 source put it
    openai.com First shared 2026-11-06
  8. Shared by 3

    Staff at Commerce's $10 million-a-year AI safety center were asked to report informally to a White House office

    “The U.S. government's AI safety testing center began informally reporting to a White House office, AP reported, and it still has no permanent leader.” via The Neuron

    Reported byThe Neuron, AI Breakfast

    Discussed onTechmeme

    How 1 source put it
    • The Neuron2026-11-09issue ↗ The U.S. government's AI safety testing center began informally reporting to a White House office, AP reported, and it still has no permanent leader.
    • AI Breakfast2026-11-09issue ↗
    apnews.com First shared 2026-11-08
  9. Shared by 3

    Introducing Arete-27B: Qwen3.8-27B Post-Trained To Reinforce Western Values

    Nous Research post-trained Qwen3.8-27B into Arete-27B to reduce CCP-aligned propaganda from 22% to 1%. Includes an open evaluation of 803 prompts, training recipes, and a dataset of twenty thousand prompts. cited

    Reported byAlphaSignal, The Neuron

    Discussed onTechmeme

    AboutQwen3.8-27B

    Where it was shared
    nousresearch.com announcement · 8 min read First shared 2026-11-06
  10. Shared by 1

    Release 0.31

    “Release: llm-anthropic 0.31 Anthropic released Fable 5.2 and Mythos 5.2 today. Fable 5.2 is currently ranked first on Artificial Analysis. Anthropic say it "solved two open problems in the physics of black holes", so clearly their internal benchmarks have become more ambitious! I've added the new models to llm-anthropic, so you can now run prompts through…” via Simon Willison

    Reported bySimon Willison

    github.com 1 min read First shared 2026-11-09
  11. Shared by 1

    simonw/beep-locator: Locate annoying beeps

    This Python tool turns laptops and phones into a microphone array to triangulate the source of mysterious beeps. Includes a browser-based interface with remote code updates, a simulation mode, and commands to calibrate devices and view audio clips. cited

    Reported bySimon Willison

    github.com repo · 3 min read First shared 2026-11-09
  12. Shared by 1

    BeepBeep-cameraready.dvi

    A 2007 system achieves around one or two centimeters accuracy within a range of more than ten meters. Details how sample counting avoids clock synchronization errors to reach 0.8cm granularity at 44.1KHz. cited

    Reported bySimon Willison

    cs.purdue.edu paper · 57 min read First shared 2026-11-09
  13. Shared by 1

    beep-locator/beep_locator/modules/listener/detector.js at main

    This detector.js module uses FFT and a high-pass filter to identify tonal beeps from microphone audio. Includes configurable thresholds for 12 dB and 800 Hz band limits, plus parabolic peak interpolation for frequency. cited

    Reported bySimon Willison

    github.com repo · 5 min read First shared 2026-11-09
  14. Shared by 1

    Here's that shared starter session, which began with this prompt against a newly created empty GitHub repository

    “I started this project using Claude Code Cloud via my iPhone, with Opus 5.5 running in Extra thinking effort mode. Here's that shared starter session, which began with this prompt against a newly created empty GitHub repository:” via Simon Willison

    Reported bySimon Willison

    claude.ai First shared 2026-11-09
  15. Shared by 1

    beep-locator/problem.md at main

    A Python web app uses phones and laptops to locate beeping devices via acoustic triangulation. Details the BeepBeep algorithm that cancels latency errors and requires no shared clock for self-calibration. cited

    Reported bySimon Willison

    github.com guide · 9 min read First shared 2026-11-09
  16. Shared by 1

    Commits · simonw/beep-locator

    Commits on Nov 9, 2026 add selftest procedure, local ctl defaults and CLAUDE.md to the repository. Includes prior commits building the server, browser listeners, calibration, and beep location functionality. cited

    Reported bySimon Willison

    github.com repo · 2 min read First shared 2026-11-09
  17. Shared by 1

    beep-locator/beep_locator/static/runtime.js at 44fea65c918fb31a7d2c0e17797ee1a106860e23

    The runtime.js file defines a 781-line browser runtime managing audio input, output, and a rolling buffer. Includes a RingBuffer class handling gaps and a WebAudioIO class configuring getUserMedia with echoCancellation disabled. cited

    Reported bySimon Willison

    github.com repo · 16 min read First shared 2026-11-09
  18. Shared by 2

    A Message to Our Company

    NVIDIA CEO Jensen Huang apologizes for infidelity and pledges to cooperate with an independent investigation into national security concerns. Details his commitment to serve colleagues and shareholders while asking for privacy for his family. cited

    Reported byThe Rundown AI

    Discussed onHacker News 956 pts · 1,235 comments

    nvidianews.nvidia.com news report · 1 min read First shared 2026-11-08
  19. Shared by 2

    Megaspeed exec was recruited to spy on Nvidia CEO, according to US intelligence memo

    A US intelligence memo alleges China recruited Megaspeed executive Alice Huang to spy on Nvidia CEO Jensen Huang. Includes a spokesperson statement that the interactions were not unusual and the company does not share classified information with its customers. cited

    Reported byForward Future, AI Breakfast

    Where it was shared
    techcrunch.com news report · 1 min read First shared 2026-11-09
  20. Shared by 1

    1 billion in gross annualized revenue

    “…hly a year, Handshake’s AI training business went from launch to $1 billion in gross annualized revenue. What are frontier labs buying at that scale? What evals are models still nowhere close to human level on? Garrett Lord brought the people who build these reinforcement learning environments to show us. For the first time, we are walking through them…” via Latent Space / AINews

    Reported byLatent Space / AINews

    theinformation.com First shared 2026-11-09

The week to that day two or more sources

  1. Shared by 8

    Gemini Omni 1.2 Flash: Fast and cost-effective with longer clips

    Google makes Gemini Omni 1.2 Flash generally available at $0.07 per second for 720p video. Includes native 15-second generation, multi-turn conversational editing, and improved reference adherence for consistent characters and products. cited

    Reported byAlphaSignal, Latent Space / AINews, The Rundown AI, TLDR AI, Superhuman AI, The Neuron, AI Breakfast

    Discussed onHacker News 369 pts · 253 comments

    Where it was shared
    blog.google announcement · 4 min read First shared 2026-11-05
  2. Shared by 6

    Introducing Inkling 2 Small

    Thinking Machines releases Inkling 2 Small, a 594B MoE model with 17B active parameters that supports image inputs. Includes benchmark tables showing it scores 31 on the Artificial Analysis Intelligence Index and 1527 ELO on AA-Briefcase. cited

    Reported byAlphaSignal, Latent Space / AINews, TLDR AI, The Rundown AI, The Deep View, Superhuman AI

    AboutInkling

    Where it was shared
    thinkingmachines.ai announcement · 29 min read First shared 2026-11-05
  3. Shared by 5

    A proof of the Aaronson-Ambainis conjecture

    Physical Superintelligence used an Anthropic model to prove the Aaronson-Ambainis conjecture, a 2008 quantum computing problem. Includes the full proof accepted by Lean, details on how the system improved the result twice, and Stanford researchers' comments. cited

    Reported byThe Neuron, The Rundown AI, The Deep View, TLDR AI

    Discussed onTechmeme

    Where it was shared
    psi.inc news report · 7 min read First shared 2026-11-06
  4. Shared by 8

    Introducing Personal Finance in Claude

    Anthropic launches Personal Finance in Claude for US Pro, Max, and Ultra subscribers to analyze up to two years of read-only transaction data. Includes examples of spending breakdowns, upcoming payment alerts, and APR calculations, while noting the tool cannot move or invest money. cited

    Reported byAlphaSignal, The Rundown AI, TLDR AI, Forward Future, Superhuman AI, The Deep View, The Neuron, FryAI

    Where it was shared
    claude.com announcement · 3 min read First shared 2026-11-05
  5. Shared by 9

    ChatGPT Images 2.6

    “OpenAI released ChatGPT Images 2.6 with native GIF generation from text or photo prompts” via AlphaSignal

    Reported byAlphaSignal, The Neuron, Superhuman AI, Ben's Bites, The Rundown AI, FryAI, AI Breakfast, The Deep View, Mindstream

    How 1 source put it
    openai.com First shared 2026-11-04
  6. Shared by 3

    Introducing Claude Fable 5.2 and Claude Mythos 5.2

    Anthropic released Claude Fable 5.2 and Mythos 5.2, which generate outputs up to 30% faster than the previous versions. Includes a benchmark table showing Fable 5.2 scores 49.1% on FrontierCode 2.0 Xhigh effort and details pricing of $8 per million input tokens. cited

    Reported byAlphaSignal, Simon Willison

    Discussed onHacker News 973 pts · 6 comments

    Where it was shared
    anthropic.com product page · 18 min read First shared 2026-11-09
  7. Shared by 6

    OpenAI’s new model looks for software flaws

    “OpenAI just released GPT-6 Cyber, a version of GPT-6 Astra built to find security holes and write exploits: code that takes advantage of a specific bug. OpenAI rates it “Critical” for cybersecurity, the highest risk tier in its internal framework. Not exactly a subtle job description.” via The Neuron

    Reported byThe Neuron, The Rundown AI, TLDR AI, AI Breakfast, The Deep View, FryAI

    How 1 source put it
    openai.com First shared 2026-11-06
  8. Shared by 5

    Independent Assessment of AI R&D Acceleration at Anthropic in October 2026

    “METR estimates Anthropic's model capabilities are now improving 2x faster than the March 2024 to October 2025 baseline.” via AlphaSignal

    Reported byAlphaSignal, The Rundown AI, TLDR AI, Latent Space / AINews, AI Breakfast

    How 1 source put it
    metr.org 53 min read First shared 2026-11-05
  9. Shared by 3

    The components of agent memory

    Anthropic shares how Claude agents use idle time to reorganize memory and fix errors through a process called dreaming. Includes evaluations of CL-Bench 1.0 and Pokémon Red showing how specific models like Opus 5.5 close learning loops and avoid navigation traps. cited

    Reported byThe Rundown AI, TLDR AI

    Discussed onHacker News 187 pts · 103 comments

    Where it was shared
    anthropic.com analysis · 16 min read First shared 2026-11-07
  10. Shared by 3

    Statement From NVIDIA Board of Directors

    “NVIDIA🕵️ Jensen Huang admits affair as Nvidia probes spy claimsImage source: NetThunder on LinkedInThe Rundown: Nvidia’s board of directors just ordered an independent review of reports that CEO Jensen Huang was targeted by a “foreign intelligence effort,” with Huang admitting to an extramarital affair in a letter to staff and saying, “There is no excuse…” via The Rundown AI

    Reported byThe Rundown AI, AI Breakfast

    Discussed onHacker News 490 pts · 406 comments

    How 1 source put it
    • The Rundown AI2026-11-09issue ↗ NVIDIA🕵️ Jensen Huang admits affair as Nvidia probes spy claimsImage source: NetThunder on LinkedInThe Rundown: Nvidia’s board of directors just ordered an independent review of reports that CEO Jensen Huang was targeted by a “foreign intelligence effort,” with Huang admitting to an extramarital affair in a letter to staff and saying, “There is no excuse…
    • AI Breakfast2026-11-09issue ↗
    nvidianews.nvidia.com 1 min read First shared 2026-11-08