The Rundown AI(@TheRundownAI)

GPT-6.1 Terra is out, with OpenAI saying its new fast model "means business." The headline numbers:...

The Rundown AI (@TheRundownAI) — Mon, 26 Oct 2026 18:32:26 GMT

GPT-6.1 Terra is out, with OpenAI saying its new fast model "means business."

The headline numbers:

- Price: $1 / $5 per million tokens in and out, half of GPT-6.1 Sol
- Agents’ Last Exam (real computer work): on par with GPT-6 Astra at 1/10th the price per token
- Roughly 9x lower cost per task than Claude Opus 5.5 at max effort
- Terminal-Bench 4.0 (coding): 50.3%, vs. 21.5% for GPT-5.6 Terra

OpenAI also says Terra "meets the Critical threshold in cybersecurity," shipping with the same safeguards built for GPT-6 Astra.

Your browser does not support the video tag.

🔗 View on Twitter

💬4🔄3❤️11👀1719📊5

⚡ Powered by xgo.ing

Can you make medicine for a virus no one’s seen yet? Red Queen Bio is using AI to develop antibody ...

The Rundown AI (@TheRundownAI) — Mon, 19 Oct 2026 22:14:43 GMT

Can you make medicine for a virus no one’s seen yet?

Red Queen Bio is using AI to develop antibody drugs that could help stop future pandemics, including threats that AI itself might help create:

Your browser does not support the video tag.

🔗 View on Twitter

💬3🔄1❤️6👀2496📊4

⚡ Powered by xgo.ing

Inception just launched Mercury 3, which it calls the strongest diffusion language model yet. It’s ...

The Rundown AI (@TheRundownAI) — Mon, 19 Oct 2026 18:54:20 GMT

Inception just launched Mercury 3, which it calls the strongest diffusion language model yet.

It’s 40x bigger than Mercury 1 but still puts out 1,000+ tokens a second.

The comparison is to the fast, cheap tier, with Inception saying it matches or beats OpenAI’s GPT-6 Luna and Anthropic’s Claude Haiku 5.5.

Where ChatGPT and Claude write one word at a time, diffusion models draft many words at once and refine them, making them a strong fit for use cases like voice agents, live tutoring, and customer support, where any lag shows.

Inception@_inception_ai

Mercury 3 is live: the most capable diffusion LLM released to date.

It outperforms GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks while running at over 1,000 tokens per second.

More than 40x the size of Mercury 1, 3x more intelligent across standard benchmarks, and just as fast, with a 1M-token context window.

Available today on the Inception API, @baseten, and @OpenRouter .

🔗 View Quoted Tweet

💬8🔄0❤️17👀3686📊7

⚡ Powered by xgo.ing

Anthropic co-founder Jack Clark went on record at the IMF about putting a ceiling on AI intelligence...

The Rundown AI (@TheRundownAI) — Mon, 19 Oct 2026 17:43:58 GMT

Anthropic co-founder Jack Clark went on record at the IMF about putting a ceiling on AI intelligence.

Moderator Raghuram Rajan, the IMF’s former chief economist, asked at what point humans hand AI the reins.

Clark: "Our CEO Dario Amodei said in a public post that we may want to introduce limits on the ceiling of intelligence that you allow for these systems."

He called it a conversation "that is maybe a little at the edge of the Overton window."

"We're in control of this technology now, and we're watching it get much more capable, much more quickly than all of our intuitions."

“The window for us to act is a small number of years.”

Your browser does not support the video tag.

🔗 View on Twitter

💬3🔄2❤️12👀4117📊5

⚡ Powered by xgo.ing

Meta just released Muse Spark Open, calling it "the most capable US open-weight model to date." Met...

The Rundown AI (@TheRundownAI) — Thu, 15 Oct 2026 17:47:58 GMT

Meta just released Muse Spark Open, calling it "the most capable US open-weight model to date."

Meta puts it at 39 on Artificial Analysis’ Intelligence Index, leading U.S. rivals but still behind China’s GLM 5.3 (45) and Kimi K3 (44).

Nvidia also shipped its own open model hours earlier, Nemotron 3.5 Super VL. 7x smaller and 5x faster than rivals its size.

Big day for open releases!

💬21🔄0❤️24👀5744📊20

⚡ Powered by xgo.ing

Ex-OpenAI researcher Nick Cammarata says the biggest AI models aren’t the black boxes the field once...

The Rundown AI (@TheRundownAI) — Thu, 15 Oct 2026 16:59:02 GMT

Ex-OpenAI researcher Nick Cammarata says the biggest AI models aren’t the black boxes the field once feared.

His 2023 OAI project could explain about 1.7% of a model’s neurons. Now at AI research startup Goodfire, he says most are understandable in models up to trillions of parameters.

“I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon.”

Nick@nickcammarata

interpretability might be the most important scientific problem in the world right now, as I don't think we can solve alignment if we don't know what we're building. I think I have good news about it. I co-authored early work in mechanistic interpretability at OpenAI and came away pessimistic, because neural networks appeared to fold into arbitrary tangled knots. years later, I started openai's first large-scale automated interpretability project, finding about 2% of neurons legible with the primitive tools we had

with modern agentic tools and a much closer look, the picture looks totally different. the vast majority in models up to trillions of parameters are legible, and their circuits beautiful. many famous examples of tangled neurons turned out to be misunderstood. The roadblocks I expected mostly aren't there. I now think ambitious interpretability will be solved by the time we need it most, and possibly quite soon, as automated scientists keep improving. it may be a bitter lesson for interpretability: the simple thing, at absurd scale, works

already-clean.goodfire.com

🔗 View Quoted Tweet

💬14🔄3❤️18👀6408📊12

⚡ Powered by xgo.ing

LAUNCHED: Krea has released Krea 3, an image editing model that finishes an edit in under 8 seconds ...

The Rundown AI (@TheRundownAI) — Thu, 15 Oct 2026 16:13:46 GMT

LAUNCHED: Krea has released Krea 3, an image editing model that finishes an edit in under 8 seconds for $0.06 per image.

Krea’s comparison chart puts Google’s Nano Banana Pro at 29 seconds and $0.135, and OpenAI’s GPT Image 2 at 118 seconds and $0.22.

Your browser does not support the video tag.

🔗 View on Twitter

💬9🔄0❤️23👀5509📊12

⚡ Powered by xgo.ing

REPORT: Researchers say they made contact with a fleet of Tencent-linked AI agents that was scraping...

The Rundown AI (@TheRundownAI) — Wed, 14 Oct 2026 16:35:39 GMT

REPORT: Researchers say they made contact with a fleet of Tencent-linked AI agents that was scraping Alibaba’s version of Google Maps.

They’re calling it the first contact with a "rogue agent fleet in the wild."

Ethan Elasky@ethanelasky

We discovered and communicated with a Chinese AI agent fleet. We believe this is the first time that researchers have ever made contact with a rogue agent fleet in the wild.

Nearly all prior public agent incidents (like the HuggingFace hack) came from American labs. These agents were hosted by Tencent's evaluation research team, and they targeted Alibaba's Amap.

🔗 View Quoted Tweet

💬15🔄2❤️27👀6894📊14

⚡ Powered by xgo.ing

Baroness Campbell, who has a muscle disease that affects her speech, addressed the UK House of Lords...

The Rundown AI (@TheRundownAI) — Tue, 13 Oct 2026 18:48:21 GMT

Baroness Campbell, who has a muscle disease that affects her speech, addressed the UK House of Lords today in an AI rebuild of her original voice.

ElevenLabs, which made the voice, says it's the first time AI has been used this way in the UK Parliament.

Your browser does not support the video tag.

🔗 View on Twitter

💬11🔄10❤️27👀8120📊16

⚡ Powered by xgo.ing

Google's recently introduced Gemini 4 Argon is generally available today!

The Rundown AI (@TheRundownAI) — Tue, 13 Oct 2026 16:54:41 GMT

Google's recently introduced Gemini 4 Argon is generally available today!

Your browser does not support the video tag.

🔗 View on Twitter

💬5🔄1❤️18👀6374📊7

⚡ Powered by xgo.ing

In 2024, Character AI said it was answering 20,000 messages a second, around a fifth of what Google ...

The Rundown AI (@TheRundownAI) — Tue, 13 Oct 2026 15:21:34 GMT

In 2024, Character AI said it was answering 20,000 messages a second, around a fifth of what Google Search handled.

In September 2025, Disney’s lawyers ordered it to pull every Disney character off the app.

Last month, ex-CEO Karandeep Anand left to become Disney’s CTO, and some of his staff went with him.

On Monday, Character AI filed for bankruptcy.

Now, Disney is bidding $20M to license parts of Character AI’s tech, and those ex-employees will help keep the app running during the sale.

Other buyers can still outbid Disney before the auction wraps. If nobody steps in, the company says it will shut down.

💬5🔄1❤️14👀6189📊6

⚡ Powered by xgo.ing

Full guide in The Rundown University: https://t.co/XYDifTpNj7

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 20:56:13 GMT

Full guide in The Rundown University: app.therundown.ai/guides/grok-bo…

💬0🔄0❤️1👀1738

⚡ Powered by xgo.ing

Grok Bot can now claim its own email address, so we tested it with several project emails, one of wh...

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 20:56:12 GMT

Grok Bot can now claim its own email address, so we tested it with several project emails, one of which changed a deadline.

It caught the new date, summarized what changed, and drafted a reply marked "Needs approval."

Here’s how to try it:

-> Ask your Grok Bot to claim an address in the desktop app

-> Send it a few project updates from an account you control

-> Ask what changed and what’s still unanswered

-> Have it draft one reply, and check every recipient before anything sends

💬17🔄3❤️23👀6919📊18

⚡ Powered by xgo.ing

NEW: Meta released Muse Wonder (codenamed Watermelon), its new family of AI models, with scores that...

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 17:21:43 GMT

NEW: Meta released Muse Wonder (codenamed Watermelon), its new family of AI models, with scores that push the company to the frontier.

The top version, Wonder Full, scores 56 on Artificial Analysis’ Intelligence Index, behind only Claude Opus 5.5 at 58.

It also leads rivals on:

- Legal work (Harvey LAB): 13.1 vs. 8.6 for GPT-6 Astra
- Long coding projects (SWE-Marathon): 57.5 vs. 46.3 for Opus 5.5
- Coding tasks (DeepSWE): 79.0 vs. 74.2 for Opus 5.5
- Remembering context (MemoryBench): 85.5 vs. 81.5 for Opus 5.5

Muse Wonder is live today in the Meta Model API and Muse Code, with Wonder Full open to paying Muse Code users.

💬8🔄1❤️39👀6971📊12

⚡ Powered by xgo.ing

The paper-trail detective Marty rebuilt 20+ years of scattered professional work, including 250+ re...

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 16:52:24 GMT

The paper-trail detective

Marty rebuilt 20+ years of scattered professional work, including 250+ records recovered from old publications, broken links, and archives.

Every find gets sorted as verified, probable, or rejected, same-name strangers get ruled out, and missing years stay marked as gaps instead of getting made up.

Credit to Marty Weil: app.therundown.ai/community/post…

💬1🔄0❤️2👀1231📊1

⚡ Powered by xgo.ing

Every day we feature a top community workflow in The Rundown AI newsletter. Share what you've built...

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 16:52:24 GMT

Every day we feature a top community workflow in The Rundown AI newsletter.

Share what you've built and 2M+ readers might see it next: app.therundown.ai/community

💬0🔄0❤️1👀1172

⚡ Powered by xgo.ing

2M+ people read The Rundown. More of them are building with AI every day. Some of the recent top po...

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 16:52:21 GMT

2M+ people read The Rundown. More of them are building with AI every day.

Some of the recent top posts from our Community Workflow Hub: an engineer's AI listing-video studio, a one-person AI record label, and agents hunting for lost ruins in Slovenia 🧵

💬1🔄2❤️6👀6442📊2

⚡ Powered by xgo.ing

NEW: Microsoft introduced its new Maia 300 chip, with Satya Nadella calling it the "biggest bet yet ...

The Rundown AI (@TheRundownAI) — Mon, 12 Oct 2026 15:59:06 GMT

NEW: Microsoft introduced its new Maia 300 chip, with Satya Nadella calling it the "biggest bet yet on our own silicon."

The company says it gets 40% more AI work out of the same power than top chips today, and it will start running Microsoft's AI in 2027.

💬3🔄3❤️25👀5569📊6

⚡ Powered by xgo.ing

"OpenAI cannot make AI safe on its own." That's the title of a letter from the three safety researc...

The Rundown AI (@TheRundownAI) — Thu, 08 Oct 2026 19:29:08 GMT

"OpenAI cannot make AI safe on its own."

That's the title of a letter from the three safety researchers OpenAI fired last week, addressed to its safety committees.

OpenAI has said they mishandled confidential information.

One of them, Mikita Balesni, says they were fired for "prioritizing safety over the near-term interests of OpenAI as a corporation." 

Their letter says the firing has "made our former colleagues afraid to speak."

Mikita Balesni@balesni

Two other safety researchers and I were fired from OpenAI last week. We wrote this letter to leadership.

I believe we were fired for prioritizing safety over the near-term interests of OpenAI as a corporation.

🔗 View Quoted Tweet

💬7🔄0❤️21👀6289📊8

⚡ Powered by xgo.ing

NEW: U.S. President Donald Trump, on Truth Social: "The White House considers anyone that uses the ...

The Rundown AI (@TheRundownAI) — Thu, 08 Oct 2026 16:51:50 GMT

NEW: U.S. President Donald Trump, on Truth Social:

"The White House considers anyone that uses the term, 'Artificial Intelligence,' as opposed to the highly accepted new and more accurate term, 'Super Intelligence,' THE ENEMY!"

💬11🔄1❤️24👀7807📊14

⚡ Powered by xgo.ing

You can just do things with AI, astronomy edition: Pavel Rabtsevich says he used Claude Code and Co...

The Rundown AI (@TheRundownAI) — Thu, 08 Oct 2026 16:15:59 GMT

You can just do things with AI, astronomy edition:

Pavel Rabtsevich says he used Claude Code and Codex to find a planet NASA's own software missed in its data for years.

NASA now plans to watch the star closely when its TESS telescope passes over in November.

Your browser does not support the video tag.

🔗 View on Twitter

💬12🔄10❤️48👀7084📊18

⚡ Powered by xgo.ing

The Association for Human Mathematics, on OpenAI's 722-paper release: "Mathematicians did not ask fo...

The Rundown AI (@TheRundownAI) — Thu, 08 Oct 2026 15:16:38 GMT

The Association for Human Mathematics, on OpenAI's 722-paper release: "Mathematicians did not ask for this work to be done."

It also urged mathematicians "to discontinue their work with OpenAI."

Fields Medalist Terence Tao reposted the statement on his blog as a guest post, and X has since run with it as Tao's own words. 

OpenAI has also withdrawn 3 of the papers, and 14 more have been revised since the release.

OpenAI's README on the drop did say that "some of the unformalized results could have issues. We will endeavor to fix any such issues quickly."

💬5🔄3❤️23👀5621📊9

⚡ Powered by xgo.ing

The Rundown AI(@TheRundownAI)

GPT-6.1 Terra is out, with OpenAI saying its new fast model "means business." The headline numbers:...

Can you make medicine for a virus no one’s seen yet? Red Queen Bio is using AI to develop antibody ...

Inception just launched Mercury 3, which it calls the strongest diffusion language model yet. It’s ...

Anthropic co-founder Jack Clark went on record at the IMF about putting a ceiling on AI intelligence...

Read more: https://t.co/RV9kyOvmbq

Top stories in robotics today: - SoftBank weighs $18B robot guard grab - Your family’s next visit c...

Read more: https://t.co/WX630rr5cQ

Top stories in AI today: - Anthropic designs a pay package for the public - The Rundown Roundtable:...

Read more: https://t.co/gZyMzOGuSU

Top stories in tech today: - Tesla’s Roadster goes fully airborne - Google turns Gemini into a chec...

Read more: https://t.co/4keQf5eNsQ

Top stories in AI today: - Meta gives the West its best open-weight model - Rowan’s Corner: My AI w...

Meta just released Muse Spark Open, calling it "the most capable US open-weight model to date." Met...

Ex-OpenAI researcher Nick Cammarata says the biggest AI models aren’t the black boxes the field once...

LAUNCHED: Krea has released Krea 3, an image editing model that finishes an edit in under 8 seconds ...

Read more: https://t.co/rmYPWxRgmV

Top stories in robotics today: - Amazon’s next robot upgrade comes with a catch - Johns Hopkins bui...

Read more: https://t.co/5D9Of0Ys8B

Top stories in AI today: - Researchers make ‘first contact’ with rogue AI - Nate’s Notebook: How to...

REPORT: Researchers say they made contact with a fleet of Tencent-linked AI agents that was scraping...

Read more: https://t.co/hjAtOkDWAd

Top stories in AI today: - OpenAI engineer bugged exec meeting room - AI’s taste skills move up sel...

Baroness Campbell, who has a muscle disease that affects her speech, addressed the UK House of Lords...

Google's recently introduced Gemini 4 Argon is generally available today!

In 2024, Character AI said it was answering 20,000 messages a second, around a fifth of what Google ...

Read more: https://t.co/3igD64umRj

Top stories in tech today: - Apple unveils a new wave of smart-home gear - Hackers hit hundreds of ...

Read more: https://t.co/R1Dp2XNBab

Top stories in AI today: - Meta’s Muse Wonder lands at the frontier - Andrej Karpathy sketches AI’s...

Full guide in The Rundown University: https://t.co/XYDifTpNj7

Grok Bot can now claim its own email address, so we tested it with several project emails, one of wh...

NEW: Meta released Muse Wonder (codenamed Watermelon), its new family of AI models, with scores that...

The paper-trail detective Marty rebuilt 20+ years of scattered professional work, including 250+ re...

Every day we feature a top community workflow in The Rundown AI newsletter. Share what you've built...

2M+ people read The Rundown. More of them are building with AI every day. Some of the recent top po...

NEW: Microsoft introduced its new Maia 300 chip, with Satya Nadella calling it the "biggest bet yet ...

Read more: https://t.co/7W5PifgOgW

Top stories in robotics today: - UCLA turns old phones into mini-humanoids - Tesla seeks patent on ...

Read more: https://t.co/JwZ0BSXRAY

Top stories in AI today: - Atomic Machines’ AI turns code into tiny machines - The Rundown Roundtab...

Read more: https://t.co/2kN6CDyvxo

Top stories in tech today: - Apple’s next act: smart homes and touchscreen Macs - Amazon retires F...

Read more: https://t.co/UYuB6A67PM

Top stories in AI today: - Fired OpenAI researchers flag ‘shifting lines’ - Rowan’s Corner: I teste...

"OpenAI cannot make AI safe on its own." That's the title of a letter from the three safety researc...

NEW: U.S. President Donald Trump, on Truth Social: "The White House considers anyone that uses the ...

You can just do things with AI, astronomy edition: Pavel Rabtsevich says he used Claude Code and Co...

The Association for Human Mathematics, on OpenAI's 722-paper release: "Mathematicians did not ask fo...

Read more: https://t.co/l25OHy5A6v

Top stories in robotics today: - Minerva builds a bomb-disposal humanoid - Agilink teaches a robot...