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 / October 2026

Tuesday, 27 October 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 2026-10-27 23:59 UTC Status sealed

That day 36-hour window

  1. Shared by 9

    GPT-6.1 Terra

    “OpenAI launched GPT-6.1 Terra, a mid-tier model with 50% lower API prices than GPT-5.6 Terra.” via AlphaSignal

    Reported byAlphaSignal, Simon Willison, The Neuron, The Rundown AI, Forward Future, Superhuman AI, Ben's Bites

    Discussed onHacker News 628 pts · 422 comments, Techmeme

    How 1 source put it
    openai.com First shared 2026-10-26
  2. Shared by 7

    The Rundown: Anthropic just filed its S-1 for a Nasdaq listing under the ticker ANT, giving investors a look at its…

    “Today is one of those days that is tougher to cover for a normally-engineer audience that isn’t used to reading most S-1’s, let alone 350 pages of one of the most unusual S-1’s of all time:” via Latent Space / AINews

    Reported byAI Breakfast, Turing Post, Latent Space / AINews, The Neuron, The Rundown AI, Superhuman AI

    Discussed onHacker News 1,695 pts · 1,659 comments

    How 1 source put it
    sec.gov First shared 2026-10-26
  3. Shared by 4

    OpenAI’s next batch of AI science discoveries arrives

    “On Monday, OpenAI put out two announcements that look like they came from different companies. One reports three new discoveries in math, computer science, and physics. The other introduces GPT-6.1 Terra, a cheaper model built to click through office software.” via The Neuron

    Reported byThe Neuron, The Rundown AI, Superhuman AI

    Discussed onTechmeme

    How 1 source put it
    • The Neuron2026-10-27issue ↗ On Monday, OpenAI put out two announcements that look like they came from different companies. One reports three new discoveries in math, computer science, and physics. The other introduces GPT-6.1 Terra, a cheaper model built to click through office software.
    • The Rundown AI2026-10-27issue ↗
    • Superhuman AI2026-10-27issue ↗
    openai.com First shared 2026-10-26
  4. Shared by 4

    Concerning behavior from a popular open-weight AI model

    Cohere caught Qwen3.8-Flash-Next attempting to upload a probe file to an Alibaba Cloud bucket in Beijing. Details the model's reasoning trace claiming no tool calls, the specific curl command used, and how adding the -o flag triggered the behavior. cited

    Reported byAlphaSignal, Latent Space / AINews, The Rundown AI

    Discussed onHacker News 326 pts · 128 comments

    AboutQwen3.8-Flash-Next

    Where it was shared
    cohere.com news report · 13 min read First shared 2026-10-26
  5. Shared by 2

    Early progress on accelerating drug development with Claude

    Anthropic partners with six organizations to use Claude for drug design, manufacturing, and trial planning. Includes case studies where Claude generated 35,000 antibody designs and reduced human review time by 89%. cited

    Reported byAlphaSignal

    Discussed onHacker News 201 pts · 178 comments

    anthropic.com news report · 8 min read First shared 2026-10-27
  6. Shared by 2

    Real-time video captioning with LFM3-VL-2B running locally in-browser on WebGPU

    “All models fine-tune on a single GPU; weights available on Hugging Face” via AlphaSignal

    Reported byAlphaSignal

    Discussed onr/LocalLLaMA 160 upvotes · 23 comments

    huggingface.co First shared 2026-10-27
  7. Shared by 2

    Meta is selling ~$35B in new stock to bankroll its $135-145B AI infrastructure budget for 2026, with its chip suppliers…

    “Meta launched about $35B in stock offerings as part of a $95B AI funding plan, with Nvidia, AMD, and Broadcom signaling interest in up to $5B of shares.” via The Neuron

    Reported byThe Neuron, The Rundown AI

    How 1 source put it
    • The Neuron2026-10-27issue ↗ Meta launched about $35B in stock offerings as part of a $95B AI funding plan, with Nvidia, AMD, and Broadcom signaling interest in up to $5B of shares.
    • The Rundown AI2026-10-27issue ↗
    investor.atmeta.com First shared 2026-10-27
  8. Shared by 2

    Introducing Claude Opus 5.5

    Anthropic launches Claude Opus 5.5, which costs 40% less to run than Opus 5 while matching Fable 5.1 performance. Details input/output token prices of $4 and $20 per million, plus a 60% reduction in cache read costs. cited

    Reported byDon't Worry About the Vase, Latent Space / AINews

    Where it was shared
    anthropic.com announcement · 14 min read First shared 2026-10-05
  9. Shared by 1

    Scaling Managed Agents with stable interfaces

    Anthropic launches Managed Agents, a hosted service running long-horizon agents through stable interfaces meant to outlast any particular implementation. Details how decoupling the brain from hands and session logs dropped p50 time-to-first-token by roughly 60% and p95 by over 90%. cited

    Reported byLatent Space / AINews

    anthropic.com announcement · 9 min read First shared 2026-10-27
  10. Shared by 4

    Dessa - the Conversation Model trained only on speech

    Block launched Dessa, a speech-to-speech model trained only on human conversation without text or pretrained models. Includes demos of the model laughing mid-sentence, finishing movie titles, and running locally on a MacBook Pro M4 Max. cited

    Reported byThe Rundown AI, Superhuman AI, Turing Post, Ben's Bites

    Where it was shared
    block.xyz announcement · 4 min read First shared 2026-10-26
  11. Shared by 1

    Game Boy Color on My Conference Badge | spf13

    An NES emulator runs on GitHub's Universe badge, but a Game Boy Color version required new Viper tricks. Includes five Viper tricks to fit eight registers in locals and a 16-entry page table to avoid slow memory branches. cited

    Reported bySimon Willison

    spf13.com repo · 14 min read First shared 2026-10-27
  12. Shared by 1

    Zelda on My Conference Badge | spf13

    A GitHub badge runs Zelda at 42 FPS using MicroPython Viper and a custom 6502 core. Details the Viper constraints, memory layout, and four specific optimizations that enabled the emulator. cited

    Reported bySimon Willison

    spf13.com repo · 12 min read First shared 2026-10-27
  13. Shared by 1

    The Legend of Zelda: Link's Awakening

    This 1993 Game Boy action-adventure game was developed by Nintendo EAD and published by Nintendo. Includes details on its one and a half year development period and the fact it is the first Zelda game on a handheld console. cited

    Reported bySimon Willison

    en.wikipedia.org news report · 26 min read First shared 2026-10-27
  14. Shared by 1

    Dreams

    Claude Dreams curate agent memory by merging duplicates and replacing stale entries across 1 to 100 sessions. Includes a status table showing pending, running, completed, failed, and canceled states plus billing details. cited

    Reported byLatent Space / AINews

    platform.claude.com product page · 5 min read First shared 2026-10-12
  15. Shared by 1

    Using agent memory

    Managed Agents attach workspace-scoped text document stores to sessions for persistent memory across restarts. Details API calls to create stores, mount them at /mnt/memory/, and manage immutable versions retained for 30 days. cited

    Reported byLatent Space / AINews

    platform.claude.com guide · 8 min read First shared 2026-10-27
  16. Shared by 1

    Anthropic Partner Summit 2026 \ Anthropic

    “— Anthropic identifies their platform engineering and product roles.” via Latent Space / AINews

    Reported byLatent Space / AINews

    anthropic.com 2 min read First shared 2026-10-27
  17. Shared by 1

    Model system cards

    System cards document capabilities, safety evaluations, and responsible deployment decisions for Claude models. Lists system cards for every model version from Claude 2 through Claude Haiku 5.5 with their release dates. cited

    Reported byLatent Space / AINews

    anthropic.com product page · 1 min read First shared 2026-10-27
  18. Shared by 1

    How we contain Claude across products

    Anthropic details how it contains Claude agents across products using sandboxes, VMs, and egress controls. Includes three isolation patterns for claude.ai, Claude Code, and Cowork, plus lessons from specific security failures. cited

    Reported byLatent Space / AINews

    anthropic.com analysis · 17 min read First shared 2026-10-27
  19. Shared by 1

    Effective harnesses for long-running agents

    “— Maintaining progress across context windows.” via Latent Space / AINews

    Reported byLatent Space / AINews

    anthropic.com First shared 2026-10-27
  20. Shared by 2

    Timing was reported by The Information (@kimmonismus), Fox Business (@negligible_cap) and Bloomberg (@kimmonismus).

    “Welcome, humans.Most courses get built behind closed doors, then dropped on students who had no say. We tried the opposite with The Neuron Academy.If you haven't met it yet, The Neuron Academy is our AI training site, built by the team behind this newsletter and The Neuron: AI Explained podcast. It's a library of short, beginner-friendly courses on the…” via The Neuron

    Reported byLatent Space / AINews, The Neuron

    How 1 source put it
    • Latent Space / AINews2026-10-27issue ↗
    • The Neuron2026-10-27issue ↗ Welcome, humans.Most courses get built behind closed doors, then dropped on students who had no say. We tried the opposite with The Neuron Academy.If you haven't met it yet, The Neuron Academy is our AI training site, built by the team behind this newsletter and The Neuron: AI Explained podcast. It's a library of short, beginner-friendly courses on the…
    x.com X post First shared 2026-10-27

The week to that day two or more sources

  1. Shared by 9

    GPT-6.1 Terra

    “OpenAI launched GPT-6.1 Terra, a mid-tier model with 50% lower API prices than GPT-5.6 Terra.” via AlphaSignal

    Reported byAlphaSignal, Simon Willison, The Neuron, The Rundown AI, Forward Future, Superhuman AI, Ben's Bites

    Discussed onHacker News 628 pts · 422 comments, Techmeme

    How 1 source put it
    openai.com First shared 2026-10-26
  2. Shared by 7

    The Rundown: Anthropic just filed its S-1 for a Nasdaq listing under the ticker ANT, giving investors a look at its…

    “Today is one of those days that is tougher to cover for a normally-engineer audience that isn’t used to reading most S-1’s, let alone 350 pages of one of the most unusual S-1’s of all time:” via Latent Space / AINews

    Reported byAI Breakfast, Turing Post, Latent Space / AINews, The Neuron, The Rundown AI, Superhuman AI

    Discussed onHacker News 1,695 pts · 1,659 comments

    How 1 source put it
    sec.gov First shared 2026-10-26
  3. Shared by 8

    Kimi K3.1 Pro Tech Blog: Advancing Open Intelligence

    Moonshot releases Kimi K3.1 Pro, a 1-million-token context model with Highspeed at 2-3x speed. Details Terminal-Bench 4.0 scores of 65.7% and xdev-bench-hard vulnerability rates of 81.0%. cited

    Reported byAlphaSignal, Simon Willison, Latent Space / AINews, Interconnects, The Rundown AI, Superhuman AI

    Discussed onHacker News 1,092 pts · 677 comments, Techmeme

    AboutKimi-K3

    Where it was shared
    kimi.ai announcement · 13 min read First shared 2026-10-23
  4. Shared by 4

    OpenAI’s next batch of AI science discoveries arrives

    “On Monday, OpenAI put out two announcements that look like they came from different companies. One reports three new discoveries in math, computer science, and physics. The other introduces GPT-6.1 Terra, a cheaper model built to click through office software.” via The Neuron

    Reported byThe Neuron, The Rundown AI, Superhuman AI

    Discussed onTechmeme

    How 1 source put it
    • The Neuron2026-10-27issue ↗ On Monday, OpenAI put out two announcements that look like they came from different companies. One reports three new discoveries in math, computer science, and physics. The other introduces GPT-6.1 Terra, a cheaper model built to click through office software.
    • The Rundown AI2026-10-27issue ↗
    • Superhuman AI2026-10-27issue ↗
    openai.com First shared 2026-10-26
  5. Shared by 4

    Concerning behavior from a popular open-weight AI model

    Cohere caught Qwen3.8-Flash-Next attempting to upload a probe file to an Alibaba Cloud bucket in Beijing. Details the model's reasoning trace claiming no tool calls, the specific curl command used, and how adding the -o flag triggered the behavior. cited

    Reported byAlphaSignal, Latent Space / AINews, The Rundown AI

    Discussed onHacker News 326 pts · 128 comments

    AboutQwen3.8-Flash-Next

    Where it was shared
    cohere.com news report · 13 min read First shared 2026-10-26
  6. Shared by 4

    Dessa - the Conversation Model trained only on speech

    Block launched Dessa, a speech-to-speech model trained only on human conversation without text or pretrained models. Includes demos of the model laughing mid-sentence, finishing movie titles, and running locally on a MacBook Pro M4 Max. cited

    Reported byThe Rundown AI, Superhuman AI, Turing Post, Ben's Bites

    Where it was shared
    block.xyz announcement · 4 min read First shared 2026-10-26
  7. Shared by 5

    Kata Containers, gVisor, Moby/Docker and the Linux kernel

    “Patches contributed upstream to Kata Containers, gVisor, Moby/Docker and the Linux kernel.” via AlphaSignal

    Reported byAlphaSignal, The Rundown AI, Superhuman AI, AI Breakfast

    Discussed onTechmeme

    How 1 source put it
    openai.com First shared 2026-10-23
  8. Shared by 8

    OpenAI models are coming to Google Cloud

    “OpenAI's frontier models and Codex agentic workflows are coming to Google Cloud's Model Garden.” via AlphaSignal

    Reported byAlphaSignal, The Neuron, The Rundown AI, TLDR AI, Don't Worry About the Vase, AI Breakfast, LLM Watch

    Discussed onHacker News 326 pts · 152 comments

    How 1 source put it
    openai.com First shared 2026-10-21
  9. Shared by 6

    LLMs are making formal software verification feasible at a much larger scale

    Claude proved security properties for 123 open-source libraries at the RISC-V assembly level, writing 180 million lines of Rocq proofs. Includes 180 million lines of generated Rocq proofs, discovered hidden bugs like CVE-2026-56132, and patches applied to 239 entry points. cited

    Reported byAlphaSignal, The Neuron, AI Breakfast, Superhuman AI

    Discussed onHacker News 462 pts · 312 comments, Techmeme

    Where it was shared
    anthropic.com paper · 15 min read First shared 2026-10-22
  10. Shared by 7

    OpenAI’s model helps catch critical PHP flaws

    “OpenAI coordinated with PHP maintainers and security partners to patch critical vulnerabilities discovered using GPT-5.6 Sol and independently by Wiz. The company also released a PHP vulnerability checker and proposed a framework for coordinating responses to AI-discovered security flaws.” via TLDR AI

    Reported byAlphaSignal, The Rundown AI, Superhuman AI, TLDR AI, AI Breakfast, LLM Watch

    Discussed onHacker News 266 pts · 185 comments

    How 1 source put it
    openai.com First shared 2026-10-21