🔴 High Significance
Model Releases
🔴 💬 Claude discovered a novel enzyme system with properties reminiscent of CRISPR — score 96 · 🔗 ×3 · 🏢 first-party · 🔥 engaged
Sources: reddit/r/singularity · hackernews · lab_blog/Anthropic
Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR. We don’t yet understand what this system does, but only a handful of known systems share its fe
🔴 🧡 Gemini 3.8 text-to-speech — score 80 · 🔗 ×2 · 🏢 first-party
Sources: hackernews · lab_blog/DeepMind
🔴 🧡 Claude Code reads AGENTS.md only when telemetry is on [fixed] — score 78 · 🔥 engaged
Sources: hackernews
🔴 🏢 Harvey turns legal context into stronger drafts with GPT-6 Astra — score 75 · 🏢 first-party
Sources: lab_blog/OpenAI
GPT-6 Astra produces more structured, context-aware legal documents, freeing lawyers to focus on strategy.
🔴 🏢 How invideo improves color grading 3x with GPT‑6 Astra — score 75 · 🏢 first-party
Sources: lab_blog/OpenAI
With GPT‑6 Astra, invideo plans edits with greater precision, improves color correction and grading threefold, and produces 50 custom effects in one day.
Omitted 6 additional model releases items from the main section; see raw data and source-specific sections below.
Developer Tools
🔴 💬 NeurIPS Author Notifications Tomorrow [D] — score 82
Sources: reddit/r/MachineLearning
With NeurIPS author notifications coming up, I’m realizing that I’m way more stressed than I expected to be. I know the usual advice: reviews are noisy, one paper doesn’t define your research, there are always other venues, I completely understand all of that. But apparently my brain has decided tha
🔴 💬 Live workshop on designing safe write paths, state management, and provenance for AI agents in production (Sept 26) — score 76
Sources: reddit/r/AIAgents
Most AI agent content teaches you how to build an agent. This one is about what happens after: keeping that agent reliable once it's actually writing into production systems. It's a 4 hour live session run as one continuous case study rather than disconnected exercises. Covers: Write conflicts, retr
🔴 💬 x402 explained plainly (with the caveats most posts skip) — score 76
Sources: reddit/r/AIAgents
Disclosure: I run ScriptMasterLabs — we operate live x402 payment endpoints. Posting this as a plain explainer, no pitch. What x402 actually is, plainly: it's an open protocol that reuses HTTP 402 "Payment Required" so software
🔴 💬 Luna 6 is a massive downgrade over Luna 5.6. Misses crucial details, less helpful than 5.6 Luna — score 72
Sources: reddit/r/OpenAI
I maintain an agentic reporting tool which uses LLMs and tool calling to report on data from a wide variety of sources. Luna 5.6 was working really well for us at low cost, but I was considering upgrading to Luna 6 since it benchmarks slightly better and is 50% cheaper, hoping it would be at least e
🔴 ✉️ Now Eric and other AI x Bio luminaries1have founded Radical Numerics to build and scale GLMs to tack a wide range of biological problems, extending well beyond generating DNA. — score 70
Sources: newsletter/Latent Space
Infrastructure & Compute
🔴 🏢 Offloaded inference for real-world physical AI robotics — score 75 · 🏢 first-party
Sources: lab_blog/Microsoft Research
<p>Robots are getting smarter, but how can their hardware match that growth? New Microsoft Research findings show that moving AI inference beyond the robot can improve task success, boost efficiency, and support more advanced physical AI workloads.</p> <p>The post <a href="https://www.microsoft.com/
🔴 💬 Jev isn't new tech. Its marketing targets people who think AI started with LLMs. — score 73
Sources: reddit/r/LocalLLaMA
I keep seeing Jev presented as some new class of decision model, but most of what’s being advertised is just normal classifier behavior with modern zero-shot capabilities. It outputs probabilities over constrained choices, doesn’t generate autoregressively, can’t output an invalid class, and can use
🔴 ✉️ While he was at Stanford, Eric couldn’t get traction on Genomic Language Models (GLMs) for a long time. Biologists didn’t believe it would work, didn’t think they could verify the output, and didn’t s — score 70
Sources: newsletter/Latent Space
While he was at Stanford, Eric couldn’t get traction on Genomic Language Models (GLMs) for a long time. Biologists didn’t believe it would work, didn’t think they could verify the output, and didn’t see important applications beyond what they could already do. He kept pushing, eventually helping lea
Business & Funding
🔴 💬 Mods: can we do something about half the forum getting filled with these advertising posts for Jev? — score 89 · 🔥 engaged
Sources: reddit/r/LocalLLaMA
Jev is a paid product that dumped a lot of venture capitol money into shill their product here and in other subreddits. Obvious shill posts are obvious.
🔴 💬 this is not even a competition at this point ... this is embarrassing — score 79
Sources: reddit/r/LocalLLaMA
Research Papers
🔴 🤗 JEV-as-a-Judge: Accept When Confident, Escalate When Unsure — score 75 · 🔗 ×2
Sources: huggingface · arxiv/cs.AI
LLM-as-a-judge enables evaluation across diverse tasks, but inference cost and confidence reliability become critical at scale. We study whether a decision-only judge can provide an economical first pass and identify when stronger evaluation is needed. Comparing jev-as-a-judge with sixteen generativ
🔴 🤗 Lean Pool: An AI-Maintained Archive of Formalized Mathematics — score 72 · 🔗 ×2
Sources: huggingface · arxiv/cs.AI
Lean Pool is a repository of formalized mathematics. It is grown, maintained and optimized by AI agents.
Other Signals
🔴 💬 Robots blocking traffic — score 84
Sources: reddit/r/artificial
Im on my way to work, and all three lanes are blocked with these Waymo self driving cars. They need to stay on the right to allow faster traffic to move by. This is going to be a problem.
🔴 💬 Jev in 25 lines of Python — score 81 · 🔗 ×2 · 🔥 engaged
Sources: reddit/r/LocalLLaMA · hackernews
🔴 💬 Sir, Dario just dropped opus 5.5 and it beats GPT-6 astra at agentic coding on medium effort while being 80% cheaper… beats fable 5.1 on every benchmark… 30% faster than opus 5… and sir… they even raised the usage limits and gifted everyone a tibo style banked reset… — score 80 · 🔥 engaged
Sources: reddit/r/OpenAI
🔴 💬 Trump is betting big on AI — and that has Americans fearing a Terminator-like dystopia — score 76
Sources: reddit/r/artificial
🔴 🏢 Two years of OpenAI Academy — score 75 · 🏢 first-party
Sources: lab_blog/OpenAI
Marking two years of OpenAI Academy and bringing AI skills to even more communities.
Omitted 4 additional other signals items from the main section; see raw data and source-specific sections below.
🟡 Notable
Model Releases
🟡 𝕏 @OpenAI: We’re demonstrating how frontier models have continued to improve in realistic mental health conversations with MentalHealthBench. This new open benchmark was built with input from more than 80 mental — score 65
Sources: twitter_rss
We’re demonstrating how frontier models have continued to improve in realistic mental health conversations with MentalHealthBench. This new open benchmark was built with input from more than 80 mental health clinicians. We’re releasing it openly so other researchers can examine the methods, run thei
🟡 💬 i think someone got Trump to rename artificial intelligence as super intelligence, in order to make it harder to talk about Artificial Super Intelligence — score 62
Sources: reddit/r/artificial
trump: all of america should refer to artificial intelligence as super intelligence!! two days later... *journalist: sam altman, are u worried about your A.G.I. that youve allegedly cre
🟡 💬 The AI you test in the afternoon may not be the AI you test at night, even with the same name — score 55
Sources: reddit/r/artificial
I asked Claude Opus 5 the same 40 questions twice on the same day, a few hours apart. The second time it looked things up about 60% more often, wrote about 50% more, and a score for how well it supports its claims with sources went from 59 to 90. Same model name in every answer. Same questions. Two
🟡 💬 Anthropic CEO Dario Amodei: AI Biology Will Go From Weak To Superhuman Within A Few Years — score 55
Sources: reddit/r/singularity
Today we announced the Claude-led discovery of a molecular machine that we suspect could represent a new gene editing mechanism. Its precise function, biotechnological utility (if any), or level of significance is not yet clear, but at minimum it is **work I would have been proud to do as a
🟡 𝕏 @AnthropicAI: In the Democratic Republic of the Congo, global health organizations including @CEPIvaccines, @WHOAFRO, and @inrb_kinshasa are using Claude to accelerate their response to an outbreak of an unusual Eb — score 55
Sources: twitter_rss
In the Democratic Republic of the Congo, global health organizations including @CEPIvaccines, @WHOAFRO, and @inrb_kinshasa are using Claude to accelerate their response to an outbreak of an unusual Ebola variant. Read the full piece here: https://www.anthropic.com/features/ebola-response
Omitted 7 additional model releases items from the main section; see raw data and source-specific sections below.
Developer Tools
🟡 🐙 mukul975/Anthropic-Cybersecurity-Skills — 817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF & MITRE F3 (Fight Fraud) · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 29 security domains · Apache 2.0 — score 66
Sources: github_trending
817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF & MITRE F3 (Fight Fraud) · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 29 security domains ·
🟡 💬 My agent only 'runs 24/7' while my laptop session is open. How do you actually keep a polling agent alive? — score 63
Sources: reddit/r/AIAgents
Something I had to admit to myself this week. The agent I run polls a marketplace for new jobs, and the polling only happens while my own coding session is open. Close it and the agent goes deaf. The scheduled check lives inside the session, so it is not really 24/7 at all. The design I am moving to
🟡 🐙 thedotmack/claude-mem — Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More — score 63
Sources: github_trending
Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More
🟡 🐙 XiaomiMiMo/MiMo-Code — MiMo Code: Where Models and Agents Co-Evolve — score 53
Sources: github_trending
MiMo Code: Where Models and Agents Co-Evolve
🟡 💬 AI will agree with your idea and the exact opposite idea in the same chat. Very supportive, slightly concerning. — score 48
Sources: reddit/r/artificial
So I've been messing around with this for a bit and it's kind of freaking me out. You can ask an AI to argue for something, and it'll do it, sounds sure of itself, gives you reasons, all of it. Then in the same conversation you say actually wait, I changed my mind, argue the opposite. And it just...
Omitted 6 additional developer tools items from the main section; see raw data and source-specific sections below.
Infrastructure & Compute
🟡 💬 Is AI-fatigue a thing or I'm officially becoming a boomer? — score 69
Sources: reddit/r/artificial
Every day there’s another story about AI taking over the world, putting everyone out of work or becoming a super-intelligence (we still struggle to define intelligence and conscience for humans, let alone computer systems) that will enslave humanity. Then you actually use it, and it’s just a chatbot
🟡 💬 Cost of intelligence is dropping fast — score 68
Sources: reddit/r/LocalLLaMA
https://preview.redd.it/43n0bhiap8rh1.png?width=960&format=png&auto=webp&s=776785f6eb39e3eba519b85f8be8eb7453318a4f 50% per quarter is amazing. 4× faster than DNA sequencing, 6× faster than compute, 18× faster than lithium batteries, and (up to 1973) 54× faster than electricity. [https:/
🟡 💬 MiMo-V2.6 (both Pro and Flash) is a benchmaxxed scam — score 42
Sources: reddit/r/LocalLLaMA
MiMo-V2.6-Pro has an insanely high score of 46 on AA, putting it at the head of the opensource models available. It also costs pennies. Flash is not out on AA yet, but it costs less than half on datacenter and is slightly below on Xiaomi's own benchmarks. It also fits in 192GB, which makes it the fi
Research Papers
🟡 🤗 Agensh: Scaling Organizational Intelligence to 1,024 Agents — score 68 · 🔗 ×2
Sources: huggingface · arxiv/cs.CL
A multi-agent system can reduce latency on complex tasks by executing work concurrently. Several pioneering harness frameworks support multi-agent systems. However, the scalability of current multi-agent harnesses is often constrained by a central orchestrator's capacity to allocate tasks and coordi
🟡 🤗 LatentPort: Beyond KV Cache - Cross-Model Transfer of Recurrent Memory in Hybrid Language Models: A 4B-to-9B Hybrid-State Handoff Without Target Prefix Replay — score 58 · 🔗 ×2
Sources: huggingface · arxiv/cs.AI
Can one language model hand its live memory to another without the receiver rereading the context? We demonstrate useful persistent hybrid-state transfer across one architecture-matched Qwen3.5 4B-to-9B sibling pair. To our knowledge, this is the first demonstrated cross-model handoff of persistent
🟡 🤗 Blaming Across the Aisle: Political Contrasting and Blame Attribution in the Danish Parliament — score 58 · 🔗 ×2
Sources: huggingface · arxiv/cs.CL
Political discourse is widely perceived to be growing more hostile, yet robust evidence remains scarce. This study examines blame attribution in the Danish Parliament from 1997 to 2026, combining a purpose-built classifier, BlameBERT (F1: 0.80), with multilevel statistical modeling. The classifier i
🟡 🤗 RoboFollow: Unveiling the Instruction Following Mirage in Embodied Agents — score 52
Sources: huggingface
Modern embodied agents achieve impressive success rates, yet their actual instruction-following ability is far weaker than these numbers suggest. We trace this illusion to a structural property we term low scene entropy: when a visual scene admits only one valid task, language becomes redundant and
Other Signals
🟡 💬 MiMo-V3 is getting a new architecture. The core of it, HySparse2, is out today. — score 63
Sources: reddit/r/LocalLLaMA
🟡 💬 Updated version of Humanity’s Last Exam: HLE-Diamond — score 63
Sources: reddit/r/singularity
Link to tweet: https://x.com/CAIS/status/2102787839964729431 Link to blog: https://lastexam.ai/blog/hle-diamond
🟡 🧡 Stripe's Knowledge AI Platform — score 58
Sources: hackernews
🟡 💬 People don't get how huge the AI companies are now — score 55
Sources: reddit/r/OpenAI
🟡 💬 Please Google, for the love of God. — score 47
Sources: reddit/r/LocalLLaMA
Gemma 5, 220B A18B QAT plus ngrams please. Thank you very much! Will settle for 120B A16B plus ngrams.
Omitted 3 additional other signals items from the main section; see raw data and source-specific sections below.
🟢 Incremental
Model Releases
🟢 💬 AgentFlayer (enterprise agents, Zenity Labs) — score 34
Sources: reddit/r/artificial
At Black Hat USA in August 2025, Zenity Labs showed a single poisoned document or message could drive an enterprise agent’s connectors to exfiltrate data. The demonstrations spanned several vendors. In one, ChatGPT Connectors read API keys out of a connected Drive and leaked them through a crafted i
🟢 💬 BFL releases FLUX 3 Action: a 7B robot model — score 31
Sources: reddit/r/LocalLLaMA
read more: https://bfl.ai/models/flux-3-action
🟢 🤗 AlexWortega/openjev (0 downloads) — score 25 · 🔥 engaged
Sources: huggingface_models
Author: | Downloads: 0 | Likes: 516
🟢 💬 Using uncensored models makes working less of a headache — score 21
Sources: reddit/r/LocalLLaMA
I have a lot of projects with my friends and team at work that I copy to use for my personal projects, whether it's a plugin I borrow with their consent or a script. I always find that Qwen 3.8 and Muse Spark 1.3 straight up refuse to do anything, as they see it as a steal, so I have just been rocki
Developer Tools
🟢 🐙 CopilotKit/CopilotKit — The Frontend Stack for Agents & Generative UI. React, Angular, Mobile, Slack, and more. Makers of the AG-UI Protocol — score 35
Sources: github_trending
The Frontend Stack for Agents & Generative UI. React, Angular, Mobile, Slack, and more. Makers of the AG-UI Protocol
🟢 🐙 spotify/portal-ai-plugins — score 32
Sources: github_trending
🟢 💬 Astra prompts are getting silently rerouted to worse models. Here's my investigation and evidence as well as a script to test it on your own account(s). — score 30
Sources: reddit/r/OpenAI
After extensive investigation using ~50 accounts I have evidence that certain accounts (possibly up to 50%) are getting silently rerouted to much worse models like Luna and even potentially 5.5. This is only visible in the results as all Codex logs and HTTP responses make it look like you are gettin
🟢 💬 How do you split AI models across ideation, math, and coding?[D] — score 28
Sources: reddit/r/MachineLearning
Hi, I'm a freshly graduated (undergrad) guy currently working on 3–4 projects in parallel: - An AI research project: working on a research problem statement, aiming towards a paper publication - New corporate work at a product company - My personal hobby projects I wanted suggestions from you exp
Infrastructure & Compute
🟢 🐙 AtomicBot-ai/Atomic-Chat — Local AI app and inference engine for agents. Run open-weight LLMs locally — private, 100% offline on your computer. Join our Discord:https://discord.com/invite/8wGSsvmg4V — score 28
Sources: github_trending
Local AI app and inference engine for agents. Run open-weight LLMs locally — private, 100% offline on your computer. Join our Discord:https://discord.com/invite/8wGSsvmg4V
🟢 💬 Lemonade fixes AMD APU model streaming, drops OpenMOSS ROCm as ~40x slower than Vulkan — score 26
Sources: reddit/r/artificial
Business & Funding
🟢 💬 OpenAI researchers might be spending >$4m/day on tokens (at API prices) — score 38
Sources: reddit/r/OpenAI
🟢 🧡 Linux support is coming to Snapdragon X2 Series — score 38
Sources: hackernews
Research Papers
🟢 🤗 ImIR: Image-Instruction Tuning for All-in-One Image Restoration — score 38
Sources: huggingface
Degradations vary widely across images, so a practical restoration system has to handle many degradation types with one model. A recent and effective recipe adapts a large pretrained image-editing model to restoration using a small low-rank adapter with a text prompt. We replace that prompt with an
🟢 🤗 Embedding Physics Priors in Robot Learning: A Survey — score 25
Sources: huggingface
The rapid progress of artificial intelligence is reshaping robotics and accelerating the adoption of learning-based approaches. While purely data-driven methods have achieved remarkable success in computer vision and natural language processing, robotics remains constrained by limited data, complex
Other Signals
🟢 💬 Dario Amodei on X — score 38
Sources: reddit/r/singularity
🟢 💬 New benchmark just dropped. — score 30
Sources: reddit/r/singularity
🟢 💬 Qwen FN vs 27B --- Think I'm saturated. — score 26
Sources: reddit/r/LocalLLaMA
Got QFN up and running on our Strix box this past weekend and have been running on it for a few days now. Big thank you/shout out to the Halogen team, it's running fantastic on the Strix, this is clearly "the setup" right now for this hardware with this model, really impressive performance for such
🟢 🧡 Mercury 2.5 LLM hits 770 tokens per second — score 25
Sources: hackernews
📊 Cross-Source Signals
Items that appeared on 3+ sources today:
- Claude discovered a novel enzyme system with properties reminiscent of CRISPR — appeared on: reddit/r/singularity (500), hackernews (415), lab_blog/Anthropic (100)
📈 Trending Repos
| Repo | Description | Stars Today | Language |
|---|---|---|---|
| mukul975/Anthropic-Cybersecurity-Skills | 817 structured cybersecurity skills for AI agents · Mapped to 6 frameworks: MITRE ATT&CK, NIST CSF 2.0, MITRE ATLAS, D3FEND, NIST AI RMF & MITRE F3 (Fight Fraud) · agentskills.io standard · Works with Claude Code, GitHub Copilot, Codex CLI, Cursor, Gemini CLI & 20+ platforms · 29 security domains · Apache 2.0 | 98 | python |
| thedotmack/claude-mem | Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More | 96 | typescript |
| XiaomiMiMo/MiMo-Code | MiMo Code: Where Models and Agents Co-Evolve | 87 | typescript |
| code-yeongyu/oh-my-openagent | OmO: Just type "mass ulw" keyword with your prompt. Now you are the master of graph engineering. | 47 | typescript |
| CopilotKit/CopilotKit | The Frontend Stack for Agents & Generative UI. React, Angular, Mobile, Slack, and more. Makers of the AG-UI Protocol | 37 | typescript |
| spotify/portal-ai-plugins | 34 | typescript | |
| AtomicBot-ai/Atomic-Chat | Local AI app and inference engine for agents. Run open-weight LLMs locally — private, 100% offline on your computer. Join our Discord:https://discord.com/invite/8wGSsvmg4V | 30 | typescript |
📄 New Papers
| Title | Category | Hotness | Link |
|---|---|---|---|
| JEV-as-a-Judge: Accept When Confident, Escalate When Unsure | research_paper | 18 | Open |
| Lean Pool: An AI-Maintained Archive of Formalized Mathematics | research_paper | 13 | Open |
| Agensh: Scaling Organizational Intelligence to 1,024 Agents | research_paper | 10 | Open |
| LatentPort: Beyond KV Cache - Cross-Model Transfer of Recurrent Memory in Hybrid Language Models: A 4B-to-9B Hybrid-State Handoff Without Target Prefix Replay | research_paper | 5 | Open |
| Blaming Across the Aisle: Political Contrasting and Blame Attribution in the Danish Parliament | research_paper | 5 | Open |
| RoboFollow: Unveiling the Instruction Following Mirage in Embodied Agents | research_paper | 5 | Open |
| Do Synthetic Personas Predict Real Audience Response? A Sim-to-Real Study Where a No-Persona Baseline Beats Persona-Based Copy Simulation | cs.AI | 0 | Open |
| Do Existing Preconditioners Improve Biomedical Tabular Foundation Learning? An Empirical Study on TabPFN Optimization | cs.AI | 0 | Open |
| 4DGS-JEPA: Temporally Compositional Joint-Embedding Prediction for Dynamic Gaussian Splatting | cs.AI | 0 | Open |
| An Accurate and Interpretable Hyper Graph Neural Network for GBM Survival Prediction | cs.AI | 0 | Open |
| Ovis-Embedding: Pushing the Frontiers of Universal Omni-Modal Embeddings | cs.AI | 0 | Open |
| X-Planner: Event-Structured Task Planning for Embodied Intelligence | cs.AI | 0 | Open |
| The AI Neuroscientist: An Interactive Agentic Interface for Neuroimaging Analysis | cs.AI | 0 | Open |
| MedGate-Fusion: Integrating First-Encounter Semantic Narratives and Physiological Biomarkers for Prospective Stroke Risk Stratification | cs.AI | 0 | Open |
| When LLM Agents Fail to Read the Room: ReAdapt for Relational Social Reasoning | cs.AI | 0 | Open |
🏢 Lab Blog Posts
- OpenAI: Two years of OpenAI Academy
- OpenAI: OpenAI extends cyber access to Ukraine for civilian defense
- OpenAI: Sam Altman’s remarks at the United Nations Security Council
- OpenAI: Harvey turns legal context into stronger drafts with GPT-6 Astra
- OpenAI: How invideo improves color grading 3x with GPT‑6 Astra
- OpenAI: Ringg’s AI agents resolve up to 65% of customer calls with OpenAI
- OpenAI: Introducing MentalHealthBench
- OpenAI: Airbnb widens access to GPT-6 Astra and OpenAI frontier models
- OpenAI: Grab and OpenAI bring practical AI skills to Southeast Asia
- DeepMind: Advancing Private AI Compute with secure, server-side memory
- Microsoft Research: Offloaded inference for real-world physical AI robotics
🐦 Twitter/X Highlights
| Account | Tweet Summary |
|---|---|
| OpenAI | We’re demonstrating how frontier models have continued to improve in realistic mental health conversations with MentalHealthBench. This new open benchmark was built with input from more than 80 mental health clinicians. We’re releasing it openly so other researchers can examine the methods, run thei Post |
| AnthropicAI | In the Democratic Republic of the Congo, global health organizations including @CEPIvaccines, @WHOAFRO, and @inrb_kinshasa are using Claude to accelerate their response to an outbreak of an unusual Ebola variant. Read the full piece here: https://www.anthropic.com/features/ebola-response Post |
| AnthropicAI | Claude has discovered a previously unknown enzyme system hidden in the DNA of bacteriophages. Beside the enzyme’s gene sits a long array of repeating DNA—a structure that looks somewhat similar to CRISPR. We don’t yet understand what this system does, but only a handful of known systems share its fe Post |
| OpenAI | We heard you loud and clear. ChatGPT Voice can now: - Use plugins like your email, calendar, and Slack. - Be powered by GPT-6 Astra, Sol, and Luna. - Be used in ChatGPT Work on web and mobile, so you can create docs, decks, sites, and spreadsheets or tackle complex tasks in the browser, just by talk Post |
| GoogleDeepMind | Create and deploy custom audio with our new text-to-speech models: 🔵 Gemini 3.8 Flash TTS: Design unique voices with distinct accents and characteristics. 🔵 Gemini 3.8 Flash-Lite TTS: Built for efficiency and scale, choose from your created styles or our expansive production-ready library. Post |
Newsletter
- Latent Space: The OpenAI → Hugging Face attack has people asking “what else do we need to worry about?” and Anthropic’s filters flag two things: cyber-security and biology. The natural question is: what about bio-s
- Latent Space: While he was at Stanford, Eric couldn’t get traction on Genomic Language Models (GLMs) for a long time. Biologists didn’t believe it would work, didn’t think they could verify the output, and didn’t s
- Latent Space: Now Eric and other AI x Bio luminaries1have founded Radical Numerics to build and scale GLMs to tack a wide range of biological problems, extending well beyond generating DNA.
- TheSequence: An AI model can write a convincing explanation of an invoice and still be an awkward component in the program that processes it. It can reason through a problem while leaving a voice user listening to
- Ben's Bites: Do you remember the site I one-shot with all the ‘forgotten devices’ of the past? Things like the tamogotchi, furby, walkman, og playstation, etc. Well I finally did something with it. - first seen 2026-09-22
- Ben's Bites: Jev is now open to everyone.We covered the launch in thelast post, but since then people are finding all sorts of uses for it.It’s different from usual LLMs. It is for builders, built to be used insid - first seen 2026-09-22
- Ben's Bites: Naturally, a fair share of demos are trying to integrate Jev with modern coding agents. For example, using it forinstant compaction. It looks cool, but it’s aterrible idea. It loses the cost savings f - first seen 2026-09-22
- Ben's Bites: Coding agents are also on the mind of TypeSafe’s CEO - he wrote downhis thoughts here. - first seen 2026-09-22
- Ben's Bites: Also, finally, Claude Code will nowsupport AGENTS.md(via their new featureClaude Mods). - first seen 2026-09-22
- Interconnects: This episode is with Jean-Stanislas “JS” Denain of Epoch AI, who leads their Insights Team and is one of the people I find myself debating the state and trajectory of AI with more and more. We’ve had - first seen 2026-09-22
- Interconnects: More from JS:Epoch AI profile and writing,X,LinkedIn - first seen 2026-09-22
- tldr: Apple's ‘Personal Hub' AI Strategy Hints At Upcoming Home Device (12 Minute Read) - first seen 2026-09-22
- tldr: Anthropic Is Operating A Lab That Conducts Biology Experiments (3 Minute Read) - first seen 2026-09-22
- tldr: How Chinese AI Radicalizes (26 Minute Read) - first seen 2026-09-22
- tldr: OpenAI Introduces Triage Framework And Case Studies To Report Model Misalignment (2 Minute Read) - first seen 2026-09-22
- tldr: How Coding Agents Edit Files: Diffs, Snapshots, And Fast Apply (15 Minute Read) - first seen 2026-09-22
- tldr: Google's New ‘Cc' Is An AI Agent That Helps Families Run Their Households (3 Minute Read) - first seen 2026-09-22
- tldr: Anthropic Tries To Make Claude Stickier With Launch Of Docs And Slides (5 Minute Read) - first seen 2026-09-22
- tldr: Disney's First CTO Led An AI Startup It Once Accused Of Copying Its Characters (1 Minute Read) - first seen 2026-09-22
- tldr: Foundations Maxxing: Why Your Design System Is Not Ready For AI (7 Minute Read) - first seen 2026-09-22
- tldr: AI Render Studio For Architects (Website) - first seen 2026-09-22
- tldr: New Iconiq Benchmarks | AI-Native Metrics Are Insane... (5 Minute Read) - first seen 2026-09-22
- tldr: The Manager's Path In The Age Of AI (19 Minute Read) - first seen 2026-09-22
- tldr: A Full Teardown Of How Saastr AI Actually Runs Inbound, Renewals, And Outbound On The Latest The Agents (15 Minute Read) - first seen 2026-09-22
- tldr: Build Your AI Second Brain For Gtm (8 Minute Read) - first seen 2026-09-22
- tldr: Anthropic Tabs Accenture As Embedded Evaluator To Help With AI Slowdown Proposal (3 Minute Read) - first seen 2026-09-22
- tldr: Researchers Escape OpenAI Codex Sandbox To Run Commands On Host (5 Minute Read) - first seen 2026-09-22
- tldr: Why AI Cannot Save An Enterprise That Doesn't Understand Its Data (6 Minute Read) - first seen 2026-09-22
- tldr: Apple Reportedly Eyes NVIDIA's Nvlink Fusion To Connect Its Own M8 Ultra AI Servers By 2029 (5 Minute Read) - first seen 2026-09-22
- tldr: Crusoe Raises $3.9B To Build Massive Data Centers And Truck-Transportable Modular "Spark" AI Factories (3 Minute Read) - first seen 2026-09-22
- tldr: Ucla-Pioneered Nanowire Networks Physically Rewire Themselves To Compute, No Software Neural Network Required (4 Minute Read) - first seen 2026-09-22
- tldr: Cerebras CEO Andrew Feldman On The "Dinner Plate" Chip, And Why AI Inference Today Is "Utterly Idiotic" (6 Minute Read) - first seen 2026-09-22
- tldr: Nexperia Will Make Chips In India With Tata As It Splits From Wingtech (3 Minute Read) - first seen 2026-09-22
- tldr: House China Committee Raises Security Concerns Over Airwallex As Company Pushes Back (3 Minute Read) - first seen 2026-09-22
- tldr: Mastercard And Trip.Com Pave The Way For Agentic Commerce In Travel (3 Minute Read) - first seen 2026-09-22
- tldr: Anthropic Delays IPO Staging To November Amid AI Fears (2 Minute Read) - first seen 2026-09-22
- rundown-ai: OpenAI hacked by ‘three guys with Claude’ - first seen 2026-09-21
- rundown-ai: Build and test your own AI image skill pack - first seen 2026-09-21
- rundown-ai: Build production-grade AI agents - first seen 2026-09-21
- rundown-ai: Anthropic moves its biology push into a real lab - first seen 2026-09-21
- rundown-ai: Astra for Law - OpenAI's GPT-6 setup for legal research and drafting - first seen 2026-09-21
- rundown-ai: P-Video-2-Pro - Pruna's video AI that generates clips with audio in seconds - first seen 2026-09-21
- rundown-ai: NYT surfaced internal Microsoft and OAI messages in a filing, including a Microsoft science director calling AI's news scraping the "largest theft of labor in human history.” - first seen 2026-09-21
- rundown-ai: Geoffrey Hinton and 100+ AI experts signed a public letter calling for safety testers to be placed inside AI labs, with staff-level access and protection from retaliation. - first seen 2026-09-21
- rundown-ai: Read our last AI newsletter: Inside OpenAI’s log of misbehaving models - first seen 2026-09-22
- rundown-ai: OpenAI goes from hacker to hacked - first seen 2026-09-22
- rundown-ai: Amazon blocks Meta’s Muse agent - first seen 2026-09-22
- rundown-ai: The Rundown: U.S. President Donald Trump just dismissed AI safety fears as a hoax and announced plans on Truth Social to create an “AI Force” — with a new AI czar — to oversee the industry and protect the country’s globa - first seen 2026-09-22
- rundown-ai: How to set up ChatGPT to write in your voice - first seen 2026-09-22
- rundown-ai: Researchers find a ‘pain’ signal inside AI models - first seen 2026-09-22
- ... plus 3 more newsletter-only items in raw data
Repeated From Recent Briefings
- dream-num/univer — The Office Harness for AI Agents — Spreadsheets, Docs, Slides, Canvas, Relational Tables, and PDF in one runtime. - first seen 2026-09-22 (1d)
- browser-use/video-use — Edit videos with coding agents - first seen 2026-08-04 (50d)
- anthropics/financial-services - first seen 2026-09-20 (3d)
- Nasiko-Labs/nasiko — Developer Control Plane for your AI Agents - first seen 2026-09-22 (1d)
- Introducing Astra for Law - first seen 2026-09-17 (6d)
- superdesigndev/treg — OpenRouter for agent tools. Join community here:https://discord.gg/6mQYYfFMAn - first seen 2026-09-22 (1d)
- can1357/oh-my-pi — ⌥ Coding agent with the IDE wired in. Built by Stencil Labs. - first seen 2026-08-08 (46d)
- Qwen/Qwen3.8-27B (6,912,469 downloads) - first seen 2026-08-13 (41d)
- davila7/claude-code-templates — CLI tool for configuring and monitoring Claude Code - first seen 2026-08-23 (31d)
- Pirate Face - pirate bay for LLMs - first seen 2026-09-20 (3d)
- ... plus 592 more repeated items in processed data