Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Links / Past days

Thursday, 8 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 2026-10-08 23:59 UTC Status sealed

That day 36-hour window

  1. Shared by 10

    Introducing Claude Haiku 5.5

    Anthropic launches Claude Haiku 5.5, a model that costs around 75% less to run than Haiku 4.5. Includes benchmark tables showing Haiku 5.5 scores of 46.4% on FrontierCode 1.1 and pricing details for prompts up to 100,000 tokens. cited

    Reported byAlphaSignal, Simon Willison, The Rundown AI, The Neuron, TLDR AI, Forward Future, Ben's Bites, Don't Worry About the Vase, Latent Space / AINews

    Discussed onHacker News 1,033 pts · 482 comments

    Where it was shared
    anthropic.com announcement · 5 min read First shared 2026-10-07
  2. Shared by 7

    GPT-6 and Intelligent UI for everyone

    “GPT-6 rolls out to all ChatGPT tiers with new Intelligent UI capability” via AlphaSignal

    Reported byAlphaSignal, The Rundown AI, The Neuron, TLDR AI, Ben's Bites, Don't Worry About the Vase

    Discussed onHacker News 742 pts · 447 comments

    How 1 source put it
    openai.com First shared 2026-10-07
  3. Shared by 3

    Playground

    Google Labs launches Playground, an experimental platform for creating custom games in minutes without coding. Includes single-player titles like Snooze Jump, Truth Ninja, and Polygon Pazer with specific mechanics such as 3v3 team deathmatch. cited

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

    Where it was shared
    labs.google product page · 2 min read First shared 2026-10-08
  4. Shared by 10

    Sharing AI progress in mathematics

    “a pile of mathematical research from an unreleased frontier model. These are actual research results.” via The Neuron

    Reported bySimon Willison, Latent Space / AINews, The Neuron, The Rundown AI, TLDR AI, AI Breakfast, Data Points, Don't Worry About the Vase, Ben's Bites

    Discussed onHacker News 1,321 pts · 1,503 comments

    How 1 source put it
    openai.com First shared 2026-10-06
  5. Shared by 2

    The firings: Tomek Korbak, Mikita Balesni and Jasmine Wang say OpenAI fired them last week.

    Reported byLatent Space / AINews

    Discussed onTechmeme

    x.com X post First shared 2026-10-08
  6. Shared by 2

    We're making it easier to identify AI-generated content globally.

    Google launches a global SynthID Detector for English to verify AI content from partners like OpenAI and NVIDIA. Details how the tool scans media for imperceptible watermarks left by Google and its partners since the 2023 launch. cited

    Reported byThe Rundown AI, TLDR AI

    Where it was shared
    blog.google announcement · 1 min read First shared 2026-10-08
  7. Shared by 2

    Coming next: AA Intelligence Index v5 arrives in late October with Terminal-Bench Science and a private coding set (AA).

    “Artificial Analysis Intelligence Index v5 is scheduled to launch in late October.” via AlphaSignal

    Reported byAlphaSignal, Latent Space / AINews

    How 1 source put it
    x.com X post First shared 2026-10-08
  8. Shared by 2

    Nvidia, Oracle, CoreWeave and other AI stocks sink on OpenAI revenue report

    Discussed onHacker News 350 pts · 246 comments, Techmeme

    cnbc.com 2 min read First shared 2026-10-08
  9. Shared by 2

    Nano Banana 2.1: Ranks #4 on both T2I and Editing at $0.0336 per 1K image, half its predecessor’s price (AA).

    “Built on Gemini 3.6 Flash, Nano Banana 2.1 succeeds the deprecated gemini-3.1-flash-image. Nano Banana Pro remains available as the higher-priced tier. Artificial Analysis ranked the new model fourth overall on both its AA-Image-T2I v2.0 and AA-Image-Editing v2.0 leaderboards, three places above Nano Banana 2 on each.” via AlphaSignal

    Reported byAlphaSignal, Latent Space / AINews

    How 1 source put it
    • AlphaSignal2026-10-08issue ↗ Built on Gemini 3.6 Flash, Nano Banana 2.1 succeeds the deprecated gemini-3.1-flash-image. Nano Banana Pro remains available as the higher-priced tier. Artificial Analysis ranked the new model fourth overall on both its AA-Image-T2I v2.0 and AA-Image-Editing v2.0 leaderboards, three places above Nano Banana 2 on each.
    • Latent Space / AINews2026-10-08issue ↗
    x.com X post First shared 2026-10-08
  10. Shared by 2

    Saluki 27B: “96% of Qwen 3.8’s performance at ~1/7 the size”

    “(Activity: 454): Underdog Saluki 27B is presented as a 7.89 GB ~2-bit llama.cpp-compatible quantization of Qwen3.8-27B, compressed from ~54 GB for local/offline agentic tool use on 16 GB laptops. Underdog reports 88/120 on a Berkeley Function Calling-derived tool-use benchmark, including 47/48 on…” via Latent Space / AINews

    Reported byLatent Space / AINews

    Discussed onr/LocalLLaMA 256 upvotes · 111 comments

    reddit.com Reddit thread First shared 2026-10-08
  11. Shared by 3

    Monthly API credits for Max and Team plans

    Max and Team plans receive monthly API credits ranging from $100 to $500 pooled for building custom apps. Details eligibility requiring seven days on the plan, specific roles needed to claim, and which tools like Claude Code are excluded. cited

    Reported bySimon Willison, The Neuron, Don't Worry About the Vase

    Where it was shared
    support.claude.com announcement · 5 min read First shared 2026-10-07
  12. Shared by 1

    AI Engineer New York

    AI Engineer New York 2026 runs October 12–14 at the Sheraton New York Times Square Hotel. Includes a full speaker list featuring OpenAI, Anthropic, and Two Sigma, plus details on workshops and side events. cited

    Reported byLatent Space / AINews

    ai.engineer product page First shared 2026-09-29
  13. Shared by 1

    AI Engineer Code Summit 2026: AI Engineering Conference, Nov 10-12, San Francisco

    “AIE NYC leadership tickets will sell out tomorrow, while for SF folks, AIE CODE applications are still open for the top agentic engineers in the world.” via Latent Space / AINews

    Reported byLatent Space / AINews

    ai.engineer 2 min read First shared 2026-10-08
  14. Shared by 4

    Introducing Personal Agent Protocol

    Meta and Sierra launch an open standard for personal agents to authenticate and act on company websites and APIs. Details how the protocol uses OAuth sessions to grant read-only or write access via websites, APIs, or company agents. cited

    Reported byThe Neuron, TLDR AI, AI Breakfast, Ben's Bites

    Where it was shared
    sierra.ai announcement · 3 min read First shared 2026-10-07
  15. Shared by 3

    Grok Bot opens up routing to Claude, other models

    “Grok Bot can search and monitor X now. Also, it now uses Claude Opus 5.5 as the primary model. It’ll still delegate tasks to other models based on the complexity.” via Ben's Bites

    Reported byAI Breakfast, The Rundown AI, Ben's Bites

    How 1 source put it
    x.com X post First shared 2026-10-07
  16. Shared by 1

    AI Research | Mistral

    “…ers, with open weights expected by end of month; the linked Mistral research page contextualizes this within Mistral’s broader open-weight lineup including Mistral 7B, Mixtral sparse MoE, Pixtral, Magistral, Voxtral, and Devstral. The main technical implication raised by commenters is deployment cost: a 1T-parameter open-weight MoE would likely require…” via Latent Space / AINews

    Reported byLatent Space / AINews

    mistral.ai 1 min read First shared 2026-10-08
  17. Shared by 2

    AI-assisted bank hack (reported): A CrowdStrike report, as summarized by @AndrewCurran_, attributes last week’s attack…

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

    How 1 source put it
    • Don't Worry About the Vase2026-10-08issue ↗
    • Latent Space / AINews2026-10-08issue ↗ AI-assisted bank hack (reported): A CrowdStrike report, as summarized by @AndrewCurran_, attributes last week’s attack on South Korean banks possibly to a single person. The stack reportedly combined ARTEX, DeepSeek v4.1-Flash, GLM-5.3, Grok 4.6 and Claude Code (report).
    x.com X post First shared 2026-10-08
  18. Shared by 1

    Underdog Saluki 27B

    “…f Qwen 3.8’s performance at ~1/7 the size” (Activity: 454): Underdog Saluki 27B is presented as a 7.89 GB ~2-bit llama.cpp-compatible quantization of Qwen3.8-27B, compressed from ~54 GB for local/offline agentic tool use on 16 GB laptops. Underdog reports 88/120 on a Berkeley Function Calling-derived tool-use benchmark, including 47/48 on…” via Latent Space / AINews

    Reported byLatent Space / AINews

    underdog.ai 6 min read First shared 2026-10-08
  19. Shared by 2

    llama.cpp on the stage

    “(Activity: 1040): The image (link) shows Georgi Gerganov’s llama.cpp being featured on a Microsoft/Windows stage slide titled “llama.cpp on Windows ML”, indicating Microsoft is positioning llama.cpp as part of its local AI / Windows ML ecosystem. A commenter found the likely event recording and noted the mention was brief, but the…” via Latent Space / AINews

    Reported byLatent Space / AINews

    Discussed onr/LocalLLaMA 839 upvotes · 103 comments

    x.com X post First shared 2026-10-07
  20. Shared by 1

    AI Supercomputer for Windows

    “…ted the related hardware announcements: RTX Spark laptops and DGX Station for Windows, with the DGX Station advertised as having up to 748GB total coherent memory, including 252GB at 7.1 TB/s bandwidth (NVIDIA product page). The expected six-figure pricing led commenters to view it as technically impressive but inaccessible for typical local inference…” via Latent Space / AINews

    Reported byLatent Space / AINews

    nvidia.com 6 min read First shared 2026-10-08

The week to that day two or more sources

  1. Shared by 10

    Introducing Claude Haiku 5.5

    Anthropic launches Claude Haiku 5.5, a model that costs around 75% less to run than Haiku 4.5. Includes benchmark tables showing Haiku 5.5 scores of 46.4% on FrontierCode 1.1 and pricing details for prompts up to 100,000 tokens. cited

    Reported byAlphaSignal, Simon Willison, The Rundown AI, The Neuron, TLDR AI, Forward Future, Ben's Bites, Don't Worry About the Vase, Latent Space / AINews

    Discussed onHacker News 1,033 pts · 482 comments

    Where it was shared
    anthropic.com announcement · 5 min read First shared 2026-10-07
  2. Shared by 10

    Sharing AI progress in mathematics

    “a pile of mathematical research from an unreleased frontier model. These are actual research results.” via The Neuron

    Reported bySimon Willison, Latent Space / AINews, The Neuron, The Rundown AI, TLDR AI, AI Breakfast, Data Points, Don't Worry About the Vase, Ben's Bites

    Discussed onHacker News 1,321 pts · 1,503 comments

    How 1 source put it
    openai.com First shared 2026-10-06
  3. Shared by 7

    GPT-6 and Intelligent UI for everyone

    “GPT-6 rolls out to all ChatGPT tiers with new Intelligent UI capability” via AlphaSignal

    Reported byAlphaSignal, The Rundown AI, The Neuron, TLDR AI, Ben's Bites, Don't Worry About the Vase

    Discussed onHacker News 742 pts · 447 comments

    How 1 source put it
    openai.com First shared 2026-10-07
  4. Shared by 6

    Introducing Mistral Large 4

    Mistral launches a public preview of its 1 trillion-parameter, natively multimodal model Le Chonk. Includes cybersecurity benchmarks where it scores 82% on vulnerability patching and beats GPT-6-Astra on legal tasks. cited

    Reported bySimon Willison, The Neuron, The Rundown AI, TLDR AI, AI Breakfast

    Discussed onHacker News 522 pts · 5 comments

    Where it was shared
    mistral.ai announcement · 9 min read First shared 2026-10-06
  5. Shared by 7

    Introducing Beam: Reflection’s 501B open-weight model

    Reflection AI launches Beam, a 501B sparse MoE model with 23B active parameters trained on 23.8T tokens. Details how 10.5K NVIDIA GB300 GPUs generated over 100 million rollouts to achieve 3–4× less inference compute than Qwen 3.8-Max. cited

    Reported byTuring Post, Latent Space / AINews, The Rundown AI, The Neuron, Superhuman AI, TLDR AI

    Discussed onHacker News 551 pts · 174 comments

    Where it was shared
    reflection.ai announcement · 12 min read First shared 2026-10-05
  6. Shared by 6

    OpenAI to Add Text Watermarks

    “On Monday, OpenAI said it will start hiding an invisible signal in the text that ChatGPT and Codex (its coding assistant) write for people in the European Union. The signal lives in the AI's word choices (a pattern in which words it picks), invisible to you but findable by a special detector.” via The Neuron

    Reported byThe Rundown AI, The Neuron, TLDR AI, AI Breakfast, Data Points, Don't Worry About the Vase

    How 1 source put it
    openai.com First shared 2026-10-06
  7. Shared by 5

    EmbeddingGemma 2: an open, lightweight multimodal embedding model

    Google launches EmbeddingGemma 2, a 740M-parameter Apache 2.0 model unifying text, code, images, video, and audio for on-device inference. Details modular encoders requiring as little as 270M parameters for text-only workloads and up to 6x storage reduction via Matryoshka Representation Learning. cited

    Reported bySimon Willison, The Neuron, TLDR AI, Data Points

    Discussed onHacker News 429 pts · 46 comments

    Where it was shared
    blog.google announcement · 3 min read First shared 2026-10-06
  8. Shared by 4

    Introducing Personal Agent Protocol

    Meta and Sierra launch an open standard for personal agents to authenticate and act on company websites and APIs. Details how the protocol uses OAuth sessions to grant read-only or write access via websites, APIs, or company agents. cited

    Reported byThe Neuron, TLDR AI, AI Breakfast, Ben's Bites

    Where it was shared
    sierra.ai announcement · 3 min read First shared 2026-10-07
  9. Shared by 4

    Claude now works with Google Docs, Sheets, and Slides

    Anthropic releases a beta add-on for Google Workspace that lets Claude edit Docs, Sheets, and Slides directly. Details how the sidebar reads open files, proposes edits as cards, and allows direct formula writing or slide building. cited

    Reported byThe Rundown AI, Superhuman AI, TLDR AI, Don't Worry About the Vase

    Where it was shared
    claude.com announcement · 3 min read First shared 2026-10-07
  10. Shared by 3

    Playground

    Google Labs launches Playground, an experimental platform for creating custom games in minutes without coding. Includes single-player titles like Snooze Jump, Truth Ninja, and Polygon Pazer with specific mechanics such as 3v3 team deathmatch. cited

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

    Where it was shared
    labs.google product page · 2 min read First shared 2026-10-08