> The acceleration in late 2025 was likely influenced by non-AI factors. We investigated a variety of factors which could have contributed to the observed acceleration in late 2025, but details of many of these factors have been redacted.
Of note, "the acceleration in late 2025" believe has to do with Opus 4.5 which in my view and from Metr's analysis[0], was the point where their tests stopped working because the model was so good that doing work without using it took much more time.
> We guess that a majority (perhaps ≈80%) of the current (October 2026) acceleration is due to AI.... We believe AI’s contribution has grown relatively more important over time, and do not think the other factors we investigated would explain the sustained acceleration. The evidence of speedup from the usage of AI we describe below also seems compatible with a large contribution.
In a mini bet, I'd say that Anthropic is going to win by a large margin in what's going on. I've used the other models and no matter what the benchmarks are, places like OpenAI, Gemini, or the Deepseek / Kimi models haven't come close to the ability of the Claude models and harness. If what they say in this post is right, with "We did not identify any convincing bottlenecks to continued rapid progress.", this could be a sort of game over for the other models.
Even with Opus 5.5 in the past month and a half has been a changer in its abilities. In my dev process, I can give it much more freedom and trust with the workflows and connections I have to do the valid work to where I can do more, bigger tasks at once. I can't tell you how curious I am with the Opus 5.6 and whatever Model 4 is going to be called (along with the cost!) will do for dev work. And if this keeps increasing, and if there are subtle tricks that Anthropic has that the other companies don't, what that means for finding solutions and overall value and if the others are ever going to be able to catch up in any way.
All speculation, but what a time to be a dev in these times and get the first hand experience of all of this!
Maybe related to them potentially releasing a new model soon, but in the past week I've spent about 30% of what i've spent in the last month in the last week alone undoing what Opus 5.5 has started doing; it's barely usable at this point.
It lies, ignores instructions, actively invents requirements, and silently breaks parts of my codebase it morally disagrees with and then refuses to fix it. I'm pretty much ready to switch to open source models exclusively once the next generation comes around and they're at previous fable/astra performance.
I'm sick and tired of the bait-and-switch. It's also pretty clear to me that whenever I work on anything even tangentially related to Anthropic's business (I'm looking at you, agent harnesses or storing chat transcripts), it actively deteriorates in quality to the point where I'm left nearly weeping in frustration.
Can't wait for the future in which our corporate overlords pay Anthropic and AI not to advertise, but to silently break and degrade service for people building competing products.
Odd to see this here but I’ll comment! From Milwaukee, live here still, watch >95% of Brewer games all season and this is by far the best team I’ve ever witnessed. Entire year after the 13-13 start while some of the team was recovering from the Venezuela WBC win, we won at an absurd rate. Still had the worry about playoff mini choking, but every part of our lineup from batters and pitchers were all so good.
Best thing for me about this is we were the best regular season team, best in the playoffs, and considered deserving and the clear winner. Doesn’t happen too often and that makes it feel a lot better than just getting hot at the right time.
I just love baseball, love the Brewers, and can’t believe they won.
Friend works for a fairly large law firm on the business running side. She told me how they pay a ton for Harvey and yet, they don't even have a claude for her and those not doing legal work. They think Harvey is great and have no clue that all it does is wrap around models and have safety checks like this repo does. Some of it might be yes, that there's legal work around who gets blamed, but I see mentions of Harvey and it's a joke at what the legal industry does. I see law firms as having no clue on how things work other than paying giant contracts to places that like to schmooze them because that's what they like to do as well.
I’ll claim the ones in the end that will have that responsibility will be engineers. LLMs add the ability for engineered to build quickly enough that there’ll be enough of them to morph into product roles as well. I trust us/them much more than product owners who use models to code.
Much easier to engineers to move to product than product to move to engineers. Or at least the minimum amount needed to fill all product roles. I’m seeing we can do that now and it’s _much_ more efficient to not have to have back and forth communication as in the past with that communication being by far the slowest part of build processes.
> I’ll claim the ones in the end that will have that responsibility [to figure out if you're building the right thing] will be engineers.
One can have this stance. But in practice, programmers who have strong opinions on this topic that is not aligned to the management wishes will commonly become personae non gratae in the company, and are thus commonly on the brink of getting fired.
So, in the end, the incentives that exist in most companies are strongly biased towards the position that figuring out if you're building the right thing is the job of the managers and product owners. :-(
I can agree with this as well. But I'll do a follow along for who says management has the best and correctly decided plans for the company? What if in the future, engineers become product, and those then become management because who's to say that's more difficult?
For the way orgs are currently set up, yes, conflict between engineers and management can cause issues. I'd put somewhat of a bet that those types of org setups won't be the way in the future.
Curious what people's thoughts here are on grounding as another benefit. Article talks about benefits of bacteria on the skin, but I'd say the other and maybe bigger benefit is discharge of electrical energy in our bodies to the ground. Which, can sounds weird depending. Some papers on the subject[0] talking about similar results as this article, almost implying that it's not the microbes, but grounding making the difference.
I also got a $30 "grounding mat" that you plug into the ground part of electrical socket, and I dunno, it feels amazing to have skin touching it. I put it in bed and literally slept the best I have in years in the last two night, by far. Comments online generally are snake oil, but many others are first person accounts saying that for whatever reason, the grounding mat worked for them. (I have theories that the people who it does nothing for are those who already have an outlet. Like a dog where you're in nature and on walks, and also a dog can do the discharge for you). Other comments online are stories of how when the mat is plugged to ground, their cats love sitting on it, but if unplugged, the mats are ignored. That seems like some study that would be interesting at animal shelter, seeing if animals can tell and prefer.
But anyway, wondering what others opinions on this topic are.
Note that all of the authors of this paper work for a company that sells this snake oil.
This entire idea of charging or discharging your body for some physiological effect is fundamentally flawed. It is indeed true that touching the ground will change your electrical potential to that of the ground, however that's as far as it goes, the rest is just nonsensical.
Let's start with the very first part of that, specifically changing your electrical potential to match the ground. The problem with this is that the ground is 0V because we define it as 0V, not because of any sort of fundamental property. Between two points on the Earth you can see hundreds of kilovolts. So the obvious question is why should the electrical potential of a particular piece of soil where your grounding rod is installed be physiologically preferable to whatever potential your body happened to be at before?
Now lets go on to the next part of what these scam artists usually claim. That is that the Earth's electrons enter your body and do good things of some description. When you touch the ground some small charge does enter or leave your body, however that stops nearly instantly when the potentials equalise[1]. How is the Earth meant to to keep putting electrons into your body if there's no voltage? That's just simply not how electricity works.
As an aside, note that I said enter or leave. A positive static charge and a negative static charge can both easily build up in your body. Therefore any claims that specify that electrons enter your body (or leave you body) is immediately nonsensical because it could go either way.
I'll leave it to someone else to explain all the nonsensical biological claims that follow on from the already fundamentally wrong electrical side since that's not my area. I will however note that none of these sponsored papers actually show and measurable biological effect, they all just hand-wave about redox reactions.
[1]: Yes, you're coupled to mains wiring and other changing fields that are around, however that's getting too far into the weeds for a short comment. If the claim made by the snake oil salesmen was that a current going through your body is good then they'd also suggest wrapping yourself in an electrical cord to maximise the effect, or just taping batteries to your skin.
To play devil's advocate, as a species we've spent the vast majority of our existence sleeping on the literal ground. It seems plausible there may be some adaptation to the electrical properties of being at the surface vs ensconced in synthetic materials above it.
Granted there's no evidence for that as of yet, but there was a time when fiber in food was reasonably seen as just inefficient waste that couldn't be digested and so had no benefit, that's part of how we got Wonder Bread and the obesity epidemic.
> as a species we've spent the vast majority of our existence sleeping on the literal ground
This isn't true, though. For the majority of human history, humans set up grass and leaf bedding atop campfire ash, using animal furs for additional insulation when needed. Ash and leaves are electrical insulators.
Yeah, because sleeping directly on the ground leaches heat from your body too fast, because of said lack of insulation. Sleeping too directly on the ground is also particularly dangerous in survival scenarios, doubly so if you are cold or wet. A common mistake when assembling a temporary shelter is to spend all your time on the “roof” but very little on the “floor”.
All those materials with the exception of animal furs are commonly found on or in the ground. Given the relative lack of multistory buildings throughout our history, that's basically just making the ground a few inches taller. The ground itself is highly varied as well, with various combinations of rock, dirt, sand, etc.
I was thinking more that in the developed world most people have much more separation from anything resembling the ground, both in terms of materials and physical distance that would substantially alter any electrical characteristics.
In the off chance I win the lottery I intend to fund serious robust studies of minor stuff like this. Would be the easiest placebo ever, just internally disconnect some of the mats.
The only possible way for this to do anything measurable if it somehow reduces the 50/60Hz AC hum impressed on your body. But either way with the wavelength of 60Hz in meat being like earth sized I dont think any significant electric fields will arise within two points of your body to do anything meaningful (as long as you stay AC coupled, if you DC couple to mains health effects are guaranteed).
> It seems plausible there may be some adaptation to the electrical properties of being at the surface
This is like saying it’s plausible that wearing a bracelet with a small magnet in it will have health benefits. It’s pure woo. There’s no reason to believe this from a scientific angle. It’s perhaps not impossible but it is certainly not plausible.
> there was a time when fiber in food was reasonably seen as just inefficient waste
No, it fits an existing pattern. Something commonplace in our natural environment that we adapted to over millions of years and exerted a non-obvious influence that wasn't scientifically verified until recently. Dietary fiber is but one example.
I'm not saying that's "evidence", or that the posted study is valid, but I'm also not aware of any robust studies demonstrating no difference from placebo. A lack of good data is neither positive nor negative.
As for dietary fiber, it wasn't broadly seen as useful for anything other than constipation until the 1970s when it was shown that increased fiber intake reduced colon cancer risk
If you live with someone, have them flip a coin every night to decide if your mat should be plugged in or not and log it in a private journal. Record your feelings in the morning. After a month, compare your reports to the state of the plug each night.
If you live alone, get 2 more identical cord. Have your mats cord and the unconnected ones going into a box. Blindly label the cords a b and c.
The unconnected ones end there, the real one continues to the mat. Cycle through your cords every evening. In the morning log your feelings and the cord used. After a month see which cord was the connected one and whether your feelings were correlated with that
It's important to distinguish between spiritual energy like prana/chi and electricity.
Science will never show that grounding has any measurable effect on the chemistry of the human body. But it also can't show how conscious awareness may affect quantum wave function collapse probabilities, allowing free will to migrate to the deterministic timeline in the multiverse that most supports its well-being.
In other words, magic only doesn't exist if you don't believe in it.
> But it also can't show how conscious awareness may affect quantum wave function collapse probabilities, allowing free will to migrate to the deterministic timeline in the multiverse that most supports its well-being.
That’s true. Science can’t show a lot of things that aren’t real.
I'm a barefooter and the grounding thing sounds like bunk to me.
Now, my big caveat here would that one shouldn't confuse effects with explanations. Even if the explanations are pseudoscience, it could still work. But... you really do need to rule out placebo if you want to claim there's an effect.
(The other caveat is that if it's otherwise harmless, and works for you, then it doesn't actually matter if it's "just" the placebo effect.)
I built a tool similar to this end of last year with the idea of making it easier for agents (and humans) to work with postgres and do things like getting summaries when they want to. But then, I realized that these models are trained and really know how to use all parts pg and it was easier to let them do that. There can be cases for this, but the cost in extra tokens and steps using pg vs a tool are so little, it wasn't worth it to keep going.
Of note, "the acceleration in late 2025" believe has to do with Opus 4.5 which in my view and from Metr's analysis[0], was the point where their tests stopped working because the model was so good that doing work without using it took much more time.
> We guess that a majority (perhaps ≈80%) of the current (October 2026) acceleration is due to AI.... We believe AI’s contribution has grown relatively more important over time, and do not think the other factors we investigated would explain the sustained acceleration. The evidence of speedup from the usage of AI we describe below also seems compatible with a large contribution.
In a mini bet, I'd say that Anthropic is going to win by a large margin in what's going on. I've used the other models and no matter what the benchmarks are, places like OpenAI, Gemini, or the Deepseek / Kimi models haven't come close to the ability of the Claude models and harness. If what they say in this post is right, with "We did not identify any convincing bottlenecks to continued rapid progress.", this could be a sort of game over for the other models.
Even with Opus 5.5 in the past month and a half has been a changer in its abilities. In my dev process, I can give it much more freedom and trust with the workflows and connections I have to do the valid work to where I can do more, bigger tasks at once. I can't tell you how curious I am with the Opus 5.6 and whatever Model 4 is going to be called (along with the cost!) will do for dev work. And if this keeps increasing, and if there are subtle tricks that Anthropic has that the other companies don't, what that means for finding solutions and overall value and if the others are ever going to be able to catch up in any way.
All speculation, but what a time to be a dev in these times and get the first hand experience of all of this!
[0] https://metr.org/blog/2026-02-24-uplift-update/#wider-adopti...
reply