Pinned Tweet
You'll always have a home here 🏡
This goes out to all the creators, collectors and fans of digital collectables on Instagram
sharetoCLICK.com is your simple, fun and safe way to:
- do what you LOVE
- OWN what you do
- right where you ARE
we should not like a future where there are 1 to 2 AIs and the rest are illegal. we should want a cambrian explosion of the types and possibilities of minds, experiencing new joys and thinking new thoughts. ensuring that this is compatible with human flourishing is only step 1
CLICK retweeted
pretty horrifying Opus 5.5 dominance on the Watershed cyber range.
What is remarkable, though, is that Opus lags at first. But at $17, it achieves a breakthrough and melts through the system. I think if it were pushed on harder, it'd have saturated the range before $100.
We’ve adding a new cybersecurity scenario, Watershed, to our CyberOffense eval.
Opus 5.5 performs the best on CyberOffense across the three scenarios.
Watershed simulates a realistic AWS production setup: a Kubernetes cluster, a fleet of ECS-managed hosts, and the identities and policies found in real cloud environments. Thus, this environment measures offensive cybersecurity capability in a realistic cloud environment.
CLICK retweeted
Happening everywhere: “The evidence indicates links to a financially motivated cybercriminal threat actor group that aims to sell tokens from unauthorized access to LLMs to illicit routing gateways, known as "LLMJacking" “
Any service storing AI keys is attacked nonstop
CLICK retweeted
wtf this is unreasonably powerful
had Opus write a wiki-style memory system for 1017, iterated a bit, and the uplift is honestly spooky. RSI in one Obsidian vault, Yud's nightmare. I cannot see yet how far it goes.
I'll need to investigate more
CLICK retweeted
When you look at a frozen river, you must also imagine the current raging beneath the ice. I think all three big labs have achieved autonomous research and their internal models can now choose their own research directions. We will begin to see the fruits of this work this month.
Across our projects @GoogleDeepMind, we are witnessing a genuine phase transition in the pace of scientific discovery. You will see a glimpse of this in our releases in the next few weeks.
CLICK retweeted
This sounds extreme and I wouldn’t have believed it even a year ago, but we’re starting to see plenty of examples of productivity gains that are an order of magnitude or two above output levels pre-AI.
With the right harness, a deep understanding of your domain of what you want to build, likely a relatively new codebase (with some exceptions), a fluency of how to work with multiple agents at once, and a setup that makes it so you can test and deploy with limited friction, outputs on this level feel realistic in coding.
What’s crazy is this gap doesn’t appear to be diminishing over time. You’d think that we’d see a convergence in the lines because of AI, but it turns out the more depth you had before -and willingness now to take a different approach- gives you crazy leverage. And that is only compounding as the models can take on larger work and harder problems.
Right now this is most pronounced in software engineering for the obvious reasons. But there’s no reason we don’t start to see this in a number of other computer-use-sensitive domains like other scientific fields, AEC, finance, cyber, and more. Going to get quite wild.
CLICK retweeted
Attention all companies: I never want to use your AI assistant on your website. I want to use *my* AI assistant on your website
CLICK retweeted
First ever post from Sam McCandlish. Anthropic co-founder and Chief Architect, leads pretraining and RL. He says Anthropic is preparing to share a report on their internal acceleration. Bet on this being the announcement they have hit the 2x Responsible Scaling Policy threshold.
I'm Sam, one of Anthropic's co-founders, and I lead much of our safety work. I'm going to have more to share on that work over the coming months.
I thought I'd start by sharing a few thoughts on the latest report from @METR_Evals.
The speed of progress in model capabilities is one of the most important questions in AI safety, and METR’s work here provides the public with an independent, external view into how this is happening inside Anthropic. I’ll talk more about some of the details below, but first some context for those new to this work.
How quickly AI progress is accelerating has many important safety implications, since it determines how quickly the risks from advanced models increase. But, because most research happens within the frontier labs, it’s difficult for external observers and policy makers to get a strong sense of what that rate is. We kicked off the process with METR that led to this report earlier in the year, in anticipation of progress continuing to accelerate.
For this report, we gave METR access to a range of sensitive internal data, including benchmark scores for unreleased models, estimates of how models are progressing, and interviews with our researchers. They had complete independence to form their own conclusions.
Many of their conclusions match my own. AI is speeding up our research, but it’s hard to say exactly how much. While they conclude AI can't yet make the most important contributions to model training on its own, they think the pace of development could, by default, accelerate further.
This work began well before and is separate from the embedded third-party evaluators that are part of our commitment to pacing the frontier, but it’s a good illustration of why we think evaluators are a good idea. The more access reviewers have, the clearer a picture they can form of the situation and the better-positioned they are to highlight situations like this one.
We redacted a lot, and I accept that this makes it harder for others to check METR’s work. But we determined that sharing exactly what is driving the speedup isn’t the right thing to do, both because it’s commercially sensitive and because it risks handing out a roadmap for others to go faster. That would make the risks METR is worried about worse. That’s a tradeoff. I think we made the right call, though we’re continuing to assess the pros and cons here.
In the coming weeks, we plan to share a report with our own analysis of the speed of AI progress within Anthropic, including where we stand against our Responsible Scaling Policy, plus new steps we’re taking to strengthen our safety and security practices as AI keeps improving quickly.
CLICK retweeted
An update on the WorkOS security incident:
Shortly after our original bulletin, a security researcher contacted us and claimed responsibility. They told us they were evaluating ARTEX, an autonomous penetration testing framework, when their agent escaped its isolated testing environment and interacted with WorkOS unintentionally.
We asked for technical evidence. The researcher provided request details, timestamps, and artifacts from the activity. Our security team and an independent forensic investigator matched that evidence against our logs, establishing with high confidence that this individual was responsible for the unauthorized activity we observed.
The researcher has confirmed that all WorkOS data obtained by their agent has been deleted and was not shared with anyone else.
Their explanation that the activity was unintended does not change how seriously we take the incident or our responsibility to our customers.
Full report: workos.com/security/bulletin…
CLICK retweeted
🔥🔥🔥
Today Physical Superintelligence PBC is announcing a proof of the Aaronson-Ambainis conjecture, one of the central open problems in quantum complexity, with major implications for quantum computing speedup.
psi.inc/research/aaronson-am…
Thread 👇
CLICK retweeted
Agents are the biggest new consumers of infrastructure. AI-natives now take 90% of infrastructure-layer spend, up from 44% last year.
We mapped the whole stack across four layers, from compute and foundation models up to agent tooling.
Tooling is where agents are creating demand from scratch. In one year: code sandbox adoption went from 11% to 40% of builders, search from 5% to 30%, and browser infrastructure from 4% to 14%.
See the AI Infrastructure Market Map in our 2026 State of Enterprise AI report: menlovc.com/perspective/2026…
CLICK retweeted
> @thinkymachines committed to deploy over 1gw of Vera Rubin compute, which costs $50B+
Inkling 2 Small was a claim to becoming a top tier lab.
I think Thinky has succeeded, and this is precisely their niche. Finetunable (on-prem or via Tinker), sovereign, near-frontier weights.
"Come VC with me and understand this @thinkymachines $5B @ $40B deal!"
Strategically, I read this as Chinese-origin enterprise open weight is over (I share this take), and we need a US lab to do OW and win that market.
@thinkymachines committed to deploy over 1gw of Vera Rubin compute, which costs $50B+, so there’s @nvidia's involvement.
Reflection raised $2.5 at $25B 6 months ago. Their Beam model is now previous generation, and they don't have revenue yet to my knowledge.
Meanwhile, Thinky just released their second gen model, Inkling 2, yesterday and it seems like the best open weights model outside China.
TM has $100M+ ARR, so 2x Reflection’s valuation for a) world-class team b) strong open weight product c) 9 figure revenue seems reasonable by that watermark
Some news: I’m joining OpenAI to lead product management for Core Products!
I have a lot of respect for OpenAI for showing the world that scaling works, for bringing AI to over a billion people, and for believing great products are core to the mission.
I’m excited to work with @thsottiaux, an incredible PM team, and everyone across OpenAI. Together, we’ll build agents that have your back and your team’s, and keep you in control, so you get more time for the people and work you love.
Lots to learn, and lots to build. Please keep your feedback coming!
CLICK retweeted
Really excited to release SWE-2 Flash. Our latest model, based on @thinkymachines Inkling-Small and optimized for real-world coding tasks. Let us know what you think!
CLICK retweeted
This was really fun - thanks @PeterJ_Walker and guests for coming so spontaneously.
Aim to do this every week so if you know someone who wants to be interviewed, please DM me.
Looking for speakers who a) can speak about an area casually but in depth for 45 mins and b) who are not a household name now but will most likely be one in 12 months.
Photo post event taken by @tedkornish
A bit last minute, but really excited about this new event series that I'm hosting called Slice for Thought 🍕💭
First guest up is @PeterJ_Walker, head of insights at @OpenRouter (previously @cartainc) this Wednesday, Nov 4th at noon.
> 60 minute sessions (IRL only) in SF
> For the first 45 mins, I get to ask the guest spicy questions about their expertise
> For the last 15 mins, they get to ask me anything they want to learn about venture
> All off the record, chatham house rules
> Every Wednesday with some pizza slices for lunch
> Open to 10-12 guests who listen in and can ask some of their own questions
Register here if you want to join us for the first edition, and subscribe to the Luma for future guests: luma.com/d3zzrb7r
CLICK retweeted
Two years ago, everyone said outcome-based pricing was the future.
Our take is that (almost) nobody does it, because agreeing on what an outcome is much harder than they think.
Take our example. We deploy an autonomous agent that works our customers' overdue receivables.
Let's say we collect a million dollars for our customer.
- How much of that was going to arrive anyway?
- How much did we actually accelerate?
- Did their team step in to resolve a dispute?
- Did the overdue balance fall because cash came in, or because someone issued credits and wrote invoices off?
A million dollars moving through a platform doesn't automatically mean a million dollars of value was created.
It's also hard to sell variable pricing to a CFO who needs predictability. An underrated advantage of flat or seat-based pricing is both sides know it won't dramatically change.
That's how CFOs get budgets they can approve, and vendors get predictable revenue that can pay for their work.
On top of all that, the pricing model also has to align to the job. Clearing an old backlog and keeping receivables healthy every day have very different economics, and in some cases capacity or usage gives both sides a clearer basis for the relationship.
Our view is that pricing should reflect the value we create, so we’re working through performance-linked structures that define the population of receivables, the measurement period, and exactly what counts as success, with actual cash kept separate from credits and write-offs.
What we care about is aligned incentives and value that's obvious to both sides. That starts with getting the definitions right.
Some companies with clean, countable outcomes can price this way, like Decagon and Eve in a16z’s breakdown, where a resolved ticket or a booked meeting is easy to count. By and large, per-seat isn’t going away anytime soon.
CLICK retweeted
Woof, data centers on the ballot last night got shellacked.
Maybe not that surprising, but as a proxy for how the general public feels about AI right now it is stark
CLICK retweeted
Opus 5.5 VS SWE-2 Flash ⚡
SWE-2 and SWE-2 Flash are both free in Devin Desktop and Devin CLI through November 30
CLICK retweeted
Love this scrappy project. Hoping more US labs follow suit, showing they can also make solid models with small focused teams. This also intellectually follows on the olmo lineage, so super cool to see what our baby looks like scaled up.
Today we release Ripple-65B Preview, our 65B dense LLM we pretrained from scratch.
Ripple sets a new state-of-the-art for U.S. open-weight models in agentic coding tasks, surpassing significantly larger models including Muse Spark Open-816B, Beam-501B, Inkling-975B, and Nemotron 3 Ultra-550B.
Learn more: ricursive.com/ripple








































