Someone1234

I understand your point; but I disagree - Unlimited is possible.

It gets into a "define unlimited" discussion. You could give "unlimited" usage, but still "limit" both concurrency and throttle speed. Even if you run 24/7, they still have knobs they can twist.

Have they solved GPT5.6 SOL's propensity to over-engineer and over-complicate? You'd ask SOL to do something relatively simple, and find four single-use methods, an interface, and a factory-factory.

I actually preferred 5.6-Terra not because it is technically superior (it isn't) but because it had better instincts to NOT do this stuff.

PS - Speaking of better instincts, have they closed the UI-design gap at all? I keep a Claude subscription just because /design produces significantly higher quality UI design/UI feedback/UI refinement than anything I've seen from OpenAI.

You forgot to include whatever you're proposing instead.

"Trust me bro, Astra is better" isn't perhaps as useful as you seem to believe. I'm not even saying it is right or wrong, just that my opinion on this topic is still just one additional subjective data-point.

Only thing I wish with these benchmarks is that they would run repeat tests every couple of months. Then re-rank based on that too. We've seen a lot of performance fall-off after a couple of weeks with new releases.

For people with any kind of budget, Opus 5.5's [Medium] actually can make sense dollar per intelligence/dollar per task wise. Heck, it puts some other models to shame. [Max]'s cost is completely unhinged.

My most exciting recent release is actually 5.6 Luna, not because it is the best on any index, but the dollar per work is insane value for money. I find myself more exciting by "value" than hypothetical ceilings because I'm just not in that budget category.

I cannot tell if this is a reference to RAD in the 90s or LLMs today, and I think that kind of shows my point.

I'm going to age myself real-bad here, but I remember back when RAD ("Rapid Application Development") was the hotness of the 90s. Microsoft Access, Visual Basic (6), and even web-development itself (e.g. PHP/Wordpress).

People today likely cannot relate, but a big concern back then was identical: If the non-tech users can write their own code/and or code is easier to write, why even learn to be a programmer? What's the point?

It seems absurd now but legitimately it was exactly the same stuff. Like then, programmers will fall into two camps, those that keep up and those that fall behind. The thing with LLMs (and agentic) is that it doesn't change WHAT we do but rather HOW we do it.

So will programmers exist in the early 2000s after RAD? Well turns out, yes, and in expanded numbers. Businesses just expected more from technology, and the personnel gap never really closed. With LLMs the same will likely happen - same number of people, higher expectation.

The limit on how many programmers they'd hire was never "wants." They always wanted more than they could get. The limit was budgets/money. Now more "wants" can be fulfilled (just like post-RAD/web), but the budgets are more or less similar.

PS - Although it should be noted that the economy is not in a good place currently, and we're seeing an unusual employment environment. A lot of which "AI" is blamed on, least of all because companies can save face that way.

I flagged this and wanted to explain my reasoning.

At the time of writing, the article states something that its own link doesn't support (i.e. it is inaccurate). CBP hasn't "suspended all personal prescription importation." What they've done is removed De Minimis Exemption (i.e. duty free below $800), and that's it. Meaning you can still important personal medication before and after Oct 22nd, what has changed is that now it is subject to normal duty.

Do I think this is bad for people importing personal medication? Absolutely I do, and I oppose it. However, I don't think misstating the argument is constructive nor warranted. You could make the argument using the facts on the ground, and then re-state that in the thread title.

So the story is worthy of discussion, and interesting (even the personal medication POV). But this link and this title aren't "it." Hopefully we can get a higher quality source and title on this topic.

You aren't arguing a "contrary" to what I actually wrote above. You're arguing with a strawman Redditor, and a point adjacent to what I was posting about - experts responding to topics within their actual realm of expertise.

So I won't be addressing this, for those reasons and others.

This is commonly why, on Reddit in particular, you can get eaten alive.

Someone confident but incorrect, can often sound more convincing than someone with actual expertise. The expert must add caveats/hedge, because those are the facts on the ground, whereas the person reciting google can be entirely confident.

Of course the people judging aren't experts, so they side with confidence and simplicity. Heck, just writing shorter replies on Reddit is rewarded. Nobody reads the articles, let alone a paragraph-long reply.

That all being said though, there are limits. Sometimes LLMs on high-thinking go off on full tangents based on little, and don't have the self-awareness to bring it back.

I've been using it for several months since the developer beta came out; overall I'm extremely positive on it. One of their better releases that seemingly focused more on quality and refinements than new-features.

Siri is definitely worth using now, even if it remains still a work-in-progress/needs continued refinement. It can surprise you in good ways, but it isn't consistently good yet.

The keyboard remains unfixed, as is tradition. The same issues they've "fixed" like three or four releases now, seemingly are identically broken, and I'm lost faith they CAN fix it.

The iOS Keyboard and Search on Windows, are like the perpetually broken things that both companies have "fixed" every year for the last ten+ years and yet... It is almost farcical as this point.

Yeah; plus we know they're designing with potential future local-only Vs. hybrid Vs. cloud-only in mind depending on hardware/market changes. So having it be this flexible buys them a ton of future potential.

People have speculated that with FPGAs or even an ASIC you could take some smaller models we have today, turn the chip into essentially that model-as-hardware, and get some really compelling performance/power-usage characteristics.

At that point running a GOOD quantized Q8 model on a phone is not beyond the realm.

> It's worse for the technician in this case, because he really was being followed --- that's not unlawful but it's unnerving.

Isn't that exactly what Flock are doing to tens of thousands of people a day? So why the double-standard.

I actually agree it is unnerving, I just don't agree that one should be held to a different standard than the other.

It actually looks really nice; and I'd consider myself their target demographic conceptually. My last iPhone was a Pro for context.

But at a $2K starting price, plus likely another $200 for the storage size they really expect people to purchase, in this economy, it is so far outside of my price range it is dead to me.

Which is to say: Cool tech, but I've already dismissed it as something I want evaluate for my next phone just based on price alone.

That's maybe the real death-nail for folding phones: Price hasn't yet shifted to allow them to be mass market rather than enthusiast phones.

I've found that if you attach any pictures at all: rejected.

So what I've taken to doing is adding the business without photos (almost always accepted), and then trying to attach the photos later after that goes live (mostly works). But I agree, their process is completely broken, and I doubt they have any incentive to fix it.

I'm legitimately out of the loop; what is going on/broken with Opus 5?

One thing I've noticed and HATE, is that when you increase thinking-effort, that seemingly increases response-length. Meaning that X.High is longer than High, which is longer than Medium, etc.

Which is kind of the inverse of how people work; a really smart person can condense difficult ideas into simple[r] terms. Whereas people who struggle speak a lot but say very little.

High/X.High do seem to deliver better quality results, but it sometimes feels like needle-in-haystack extracting that from the word vomit.

I agree that the marketing copy on Claude's site remain misleading.

You really need to dive into the FAQ or help pages to understand what "20x" means. In particular the very first thing they show you is "Choose 5x or 20x more usage than Pro*" without specifying 5-hour window/session window.

Contrast that with what they bury under the fold in the FAQ:

> You get everything in Pro, plus more usage: choose from 5x or 20x the usage of Pro per 5-hour session.

Much clearer. Plus nowhere on the site that I am able to locate do they compare weekly limits for Pro, Max 5x, and Max 20x. Which I don't love - getting more weekly usage is a big deal.

I have no idea if the lawsuit will go anywhere, but regardless Anthropic need to do better. Both in the summary and in the comparison table, clearly distinguish 5-hour window limit differences from weekly total usage limit differences.

You don't see how a link to "https://claude.ai/.." in the Git log is an ad for Cluade/Anthropic? It feels exactly like "Sent from an iPhone" to me, and has exactly the same value to people reviewing the Git log later (i.e. none).

Keep in mind that if you run across someone else's spam in the Git logs:

- You need a Claude account.

- You need permission to view the session.

The only people who can just follow that link and look at how Claude was used, is Anthropic themselves. Therefore, they can also use it as an additional mechanism to tie sessions back to other IDs/identities. For everyone else this is "Sent from an iPhone" but longer.

From my point of view; I share Debian's perspective:

https://lwn.net/Articles/1091231/

Which is actually why I DON'T like the session amended. Ultimately it is an ad placed into the Git log, and noise at that. Regardless of if it exists or not, the professional responsibility for what is contained in that PR/commit falls to the feet of the developer contributing it.

Microsoft previously did this with Copilot, which was correctly met with a very negative response. I don't see why an ad by Anthropic is better than an ad from Microsoft.

From my understanding scraping a public website you'd otherwise have access to is legal in the US. However, re-hosting that scrapped copyright content, particularly if it isn't transformative/fair-use (inc. only partial content) is very much either violates copyright, or best case is legally a gray area.

Google got dragged into court for exactly this, and they were summarizing/snipping, so they had a fair-use argument to make. This is a 1:1 reproduction of the content, so the legal arguments on fair-use would be harder to make I suspect.

I've personally used XCancel, and I really like it, but stepping back legally they're on shaky ground if Twitter decides to stop it.

The EU just had to pass a law to force companies to disclose if a customer service agent is AI or Human. It is beaten.

https://commission.europa.eu/news-and-media/news/safer-and-m...

I'm genuinely concerned to ask what you're proposing?

Drug use is already unlawful, and the homeless are routinely criminalized for existing - so if that is "tolerate" then what proposals are you trying to promote?

It does kind of say: The GUID is coming from the moderation endpoint, which is hit when you generate a local or cloud AI image based on your prompt. If there is no prompt, there is no endpoint, and likely no GUID.

Obviously Paint could have been watermarking prior to AI though, but this specific AI watermarking appears to be only that.

Unfortunately they may be doubling down.

Apple is rumored to be integrating cameras into AirPods[0] that are cloud connected for "AI." They supposedly won't be able to take "pictures" and "video" but will transmit essentially the same information for AI/Siri.

Between the reaction to Flock and also the Meta Glasses; I'm legitimately surprised that companies aren't reading the Tea leaves here. But, no, I guess everything needs a camera and a cloud feed of that camera's vision now.

PS - And before you say "that is just a rumor" the video file was extracted from Apple's macOS 26.7 RC that showcases how it may work.

[0] https://www.macrumors.com/2026/08/17/camera-equipped-airpods...

Looks like in Ohio the thresholds are:

- 0 to $999: misdemeanor

- 1K: 5th-degree felony

- 7.5K-150K: 4th-degree felony

- 150K+: 3th-degree felony

But what I found interesting is - They aren't automatically inflation adjusted, it has to be manually adjusted by legislation. Which hasn't been done since 2011 meaning that in 2026 terms it would be $1.5K~ for a 5th-degree felony for like-for-like level of damage.

Kind of nuts how a poorly written law can increase the number of felonies just due to inflation alone. Although a lot of things aren't automatically inflation adjusted (e.g. minimum wage), and I think commonly it is on purpose rather than an accident. They WANT harsher sentences for relative low crimes.

Here is a fun fact; the US believes owning and wearing an F-91W is evidence that you're a terrorist so much so that they can place you in Guantanamo Bay without trial[0]:

> In a handful of cases the detainee's possession of a Casio watch or the wearing olive-drab clothing is cited as evidence that the detainee is an enemy combatant. No basis is given to explain why such evidence makes the detainee an enemy combatant.

This is also noted on the F-91W's Wikipedia page:

https://en.wikipedia.org/wiki/Casio_F-91W#Usage_in_terrorism

[0] https://en.wikipedia.org/wiki/Seton_Hall_reports#Detainees'_...

> For all the hate I have for the COVID vaccine

Can you clarify that? It saved 100K+ lives even by incredibly conservative estimates, and allowed life to returned to normal at least a year earlier than otherwise.

> result is known to have significant issues.

So well known you neglected to state or cite them? I'd invite you to do so now.

In the previous (2024) Clacton election, Conservative + Reform got a combined 74.1% of the vote, with Labour + Liberal Democrats + Greens getting a combined 24.8%.

Considering Count Binface got 26.93% (i.e. +2.13% Vs. the previous combined), the opposition did in fact converge on him. It just so happen that in Clacton the opposition is tiny, it is a very right wing area - which is why Farage picked it.

So I'd not consider +2.13% over the previous combined opposition a bad result, quite the opposite. Unless you believe previous Conservative voters would pick Binface over Farage, which to me seems highly unlikely.

All numbers from:

https://en.wikipedia.org/wiki/2026_Clacton_by-election#Resul...

The Conservative party ran in the last election, they didn't run in this election, so more right wing voters likely moved from Conservative to Farage than he got 110% of his normal core/base.