Good morning, {{ first_name | AI enthusiasts }}, and welcome to our 3,428 new readers. We’re only three days into September, and a flurry of new AI models has already hit the leaderboards.
Both Meta and Google just joined the launch party Anthropic kicked off, but on very different trajectories — Zuck and co. keep climbing into rarefied AI territory, while Google is out to prove it still has a pulse after a year to forget.
-
Meta, Google join the September AI launch party
-
Nate’s Notebook: Tech literacy = your AI ceiling
-
Nail job interviews with the “Proof Project” method
-
Report: OpenAI’s loops hit AI safety monitoring
META & GOOGLE
Image source: Meta/ Google
The Rundown: Meta and Google both just shipped new models, with Mark Zuckerberg pitching Meta’s Muse Spark 1.3 as “frontier performance almost too cheap to meter,” and Google’s Gemini 3.8 Flash showing a stronger bounce-back from its 2026 struggles.
-
Spark 1.3 (Max) comes in at a 62 on AA’s Intelligence Index, now trailing only Claude Fable 5.1 and Opus 5 despite being significantly cheaper to use.
-
Zuckerberg said “next up” is Meta’s larger model codenamed “Watermelon,” as well as releasing the weights for Muse Spark.
-
Gemini 3.8 Flash keeps 3.7’s pricing ($0.75 / $3.75) with gains in coding, reasoning, and agentic tasks, slotting in at a 59 on the Intelligence Index.
-
DeepMind’s Koray Kavukcuoglu admitted Gemini sits “a little below the frontier”, and said, “there’s nothing other than being at the frontier that’s important for us.”
Why it matters: Meta continues to climb with an impressive blend of intelligence and cost (something Google used to own), making Watermelon a potential industry-shaker. For Google, 3.8 is a step in the right direction, but with its Pro model seemingly out of the picture until Gemini 4, there is still some major pressure for a frontier comeback.
TOGETHER WITH WEIGHTS & BIASES BY COREWEAVE
The Rundown: Every enterprise racing to adopt AI eventually hits the same walls: where does the model run, where does the data go, and who can see it. Join experts from CoreWeave and Cosine on Sept. 10 for a practical session for teams deploying AI in regulated environments.
-
Define “sovereign” for your org across data residency, deployment, model provenance, and more
-
Turn governance requirements into automated checks with the open-source Responsible AI Toolkit
-
See the full sovereign AI stack come together in a live technical demo
THE RUNDOWN 101
The Rundown: Each week, Rundown AI educator Nate Grahek breaks down what he’s seeing while teaching real teams to use AI.
Nate: Two weeks ago it was Uber’s engineers. Last week it was the AI-pilled people at your small business. Notice what they have in common: they were already technical, or at least really tech-savvy. That’s the truth about the ceiling I promised.
Here’s what changed for me: I stopped nodding along. When a developer mentions a webhook or an environment variable, I don’t file it under “not my department” anymore. I turn on voice mode on the walk home and say, “Explain that to me like I run the company but not the codebase.” There are no bad questions in that room, and it will re-explain at whatever altitude you ask for.
AI raises the floor, but the ceiling stays low if you don’t build your tech literacy along with your AI fluency. And I’m talking about a pretty basic level, the way a CEO talks to their CTO. You don’t write the code, but you know what a server is, which of your files live in the cloud versus on your laptop, and how to keep them organized enough to find one.
If that’s already you, keep going. The two skills help each other immensely, and every bit of it compounds.
If you closed that door years ago (“I’m not a technical person, I never will be”), here’s the great news: there has never been a better time to open it. You now have the most incredible tech support, explainer, and duet partner sitting alongside you. With screen recording or an agent driving your browser, it walks you through things that used to require hours and hours of tech support.
So slow down and let AI teach you your own machine. What’s one term you’ve been nodding along to for years? Ask AI to explain it this week, at exactly the level you want.
AI TRAINING
The Rundown: In this guide, you will learn how to turn an AI workflow you know into a “proof project” for a job application or interview. You will record a walkthrough, use its transcript to build a five-slide deck, and tailor the finished kit to the role.
-
Pick one AI workflow you know well. Choose something simple but useful, like an AI reporting routine, research process, or content workflow
-
Use a tool like Loom to record a five-minute video of your screen and face. Walk through the AI solution, but emphasize the “human-in-the-loop” aspect
-
Paste the Loom transcript, link, and job description into ChatGPT/Claude. Tell it: “Create a five-slide deck with the video embedded on one slide. Use the other slides to explain my thought process and how the workflow operates”
Pro tip: Loom has some light editing and AI features, but the goal is to record this in one take without wasting time on edits.
TOGETHER WITH RENDER
The Rundown: Apps and agents increasingly need to run long-running background tasks — but serverless is too restrictive, and managing workers and queues by hand is a headache. Render Workflows gives you the best of both in a lightweight SDK.
With Render Workflows, you can:
-
Define tasks in a few lines of code and chain them into long-running, distributed workflows
-
Launch agents and batch jobs on demand, without spinning up infrastructure
-
Let Render handle queuing, orchestration, and retries automatically
OPENAI
Image source: Images 2.0 / The Rundown
The Rundown: OpenAI’s upcoming ‘Astra’ model expected this week reportedly uses a new technique that provides a boost to its coding and computer-use skills, according to The Information — but also keeps parts of its reasoning out of view for safety monitoring.
-
“Recurrent depth” has the model analyze the same text in repeated loops before answering, squeezing extra intelligence out of the model without making it larger.
-
The loops reportedly improve performance but come with thinking outputs that are pure math, instead of the more readable reasoning in past models.
-
OAI reportedly dialed the loops back so Astra still writes its reasoning out, and its blog promises extra monitoring of that reasoning at launch.
-
Chief Scientist Jakub Pachocki posted that he wants to “prevent a race into unmonitorability” from “confused reporting,” while calling monitoring “fragile.”
Why it matters: The alarm bells are less about Astra, which OAI says still shows its work, and more about a performance race that tempts the industry to loop harder and monitor less. Pachocki himself admits current monitoring is already fraying for reasons unrelated to Astra, so the main window into what models think may be closing either way.
COMMUNITY AI WORKFLOW OF THE DAY
▸ John built a custom GPT to explore unified physics equations
Today’s workflow comes from reader John Gaus:
“I built a custom GPT with the personalities of Einstein, Lorentz, Planck, and Compton. Starting with Einstein’s 1920 Leiden lecture, I developed an ontology for physical space based on his description of the “new ether.”
I began with E=mc2, E=hf, and my ontological modeling assumptions. I then pushed ChatGPT to challenge those assumptions and interpretations logically and mathematically. We used several tool calls and reference sites to write and test the math.
After many months of working on it in my spare time, the GPT now unifies equations that balance from the atomic scale to the black hole scale, with some remarkable revelations. I summarized much of the work in a paper written by my GPT and am happy to share it so others can expand, improve, and test the work. AI rocks.”
-
🔐 Incogni – Erase sensitive data like addresses and phone numbers from the web. Get 55% off with code RUNDOWN.*
-
⚡️ Gemini 3.8 Flash – Google’s strongest Flash model yet for coding and agents
-
🚀 Muse Spark 1.3 – Meta’s new cost-effective, near-frontier model
-
🎥 H3 Max Turbo – Fal’s video model at twice the speed, half the cost
Get more from Claude – Your style, voice, and data, supercharged. Unlock Claude’s full potential with custom skills. Download the guide*
U.S. Commerce Secretary Howard Lutnick told Axios, “We trust Anthropic,” revealing the company is “back on the right side” of the relationship with the government.
The U.S. government filed a brief in support of OpenAI in the NYT copyright lawsuit, arguing that limiting fair use for AI training would cost the U.S. its global tech lead.
New York City public schools are banning AI till eighth grade, with Mayor Zohran Mamdani rejecting that AI-powered education is “not only inevitable, but necessary.”
Spanish AI startup Multiverse Computing launched Quasar 438B, its first model, coming in at a 43 on the Artificial Analysis Intelligence Index to top all European rivals.
That’s it for today!
Before you go we’d love to know what you thought of today’s newsletter to help us improve The Rundown experience for you.
Rowan, Zach, Shubham, Jennifer, and Nate — the humans behind The Rundown
