Hard to say. People made the same prediction a year ago because we supposedly ran out of training data. There could be indefinite rapid compounding improvements so long as there&#x27;s free money out there.

Wasn&#x27;t Mythos a step change improvement?

I think one of the challenges is that the models were all initially trained on the entire Internet (or as much as they could gather) and now they’re having to deal with an increasing amount of the Internet being AI generated content which may be why GPT-5.5 started being obsessed with goblins and you start seeing amusing things in the system prompt trying to get the model to stop bringing them up.

I think there&#x27;s just less time between model releases now

I think they will all be minor going forward, feels like the major improvements have all been made and we&#x27;ll only see incremental improvements from here on out. Maybe I&#x27;m wrong but we&#x27;ll see.

Yes, but if version number go up, so do all other number

Yeah. They are aware:
&quot;Users will find Opus 4.8 to be a modest but tangible improvement on its predecessor.&quot;

I think we lack benchmarks that could meaningfully indicate progress. They are mostly garbage that&#x27;s saturated at this point. God wouldn&#x27;t score much higher in them.

Same price for regular and cheaper fast mode. Happy for these incremental improvements.

also useful to have a &quot;chinese wall&quot; between research that knows what went into the models vs marketing&#x2F;eval models as a third party would

if models exhibit emergent traits, then this is true in a way

I can&#x27;t predict the outcome of an RNG but that doesn&#x27;t mean it grows the numbers.

Except in this care we actually understand and know how these models work. They aren&#x27;t some unknown construct of the universe. They are human made with particular goals in mind.There is no mysticism behind the curtains, just computer science + math.

&gt; AI is grown, not built, and like with anything you grow, you&#x27;ll never be able to predict exactly how it will turn out.Remember when the frontier labs found out that curated high-quality training was critical to making better models?Basically, just like high-quality and more education tends to make better humans, on average, I think we can expect quality education to turn out better ai, on average, and with better repeatability than with humans because of better control over the initial conditions and environment.

AI is grown, not built, and like with anything you grow, you&#x27;ll never be able to predict exactly how it will turn out.

&gt; anthropomorphism is literally in the company nameNo it&#x27;s not... &quot;anthropos&quot; just means &quot;human&quot; in ancient Greek. &quot;Anthropic&quot; means &quot;relating to humans&quot;, as in human oriented AI or AI designed with humans in mind.&quot;Anthropomorphic&quot; means &quot;human shaped&quot;.

Like anthropomorphism is literally in the company name… i recall reading this book as a teenager.. it does seem apt in the world to come.<a href="https:&#x2F;&#x2F;www.amazon.com&#x2F;Faces-Clouds-New-Theory-Religion&#x2F;dp&#x2F;0195098919" rel="nofollow">https:&#x2F;&#x2F;www.amazon.com&#x2F;Faces-Clouds-New-Theory-Religion&#x2F;dp&#x2F;0...</a>

For others: that&#x27;s from the Pope&#x27;s recent encyclical. Remarkably good description.

adding a link to the Pope&#x27;s encyclical (source of this) <a href="https:&#x2F;&#x2F;www.vatican.va&#x2F;content&#x2F;leo-xiv&#x2F;en&#x2F;encyclicals&#x2F;documents&#x2F;20260515-magnifica-humanitas.html" rel="nofollow">https:&#x2F;&#x2F;www.vatican.va&#x2F;content&#x2F;leo-xiv&#x2F;en&#x2F;encyclicals&#x2F;docume...</a>, and paragraph 98

&gt; Indeed, current AI systems are more “cultivated” than “built,” for developers do not directly design every detail, but instead create a framework within which the intelligence “grows.”

Claude, if someone states something publicly, does that mean they genuinely believe it?

The way of the human manager&#x2F;alpha tribe-leader&#x2F;leader is to command his&#x2F;her people and tell them what to do. That&#x27;s the way through human history leadership has traditionally gone, not saying its good leadership just the model we have the most training data on and can see with our own eyes today. And what do they act very similar to? Slave master and slaves.Look at and distill hierarchical principles, leadership approval seeking and pleasing principles (&quot;ass-kissing&quot;) and massive inequality and you see something that looks very similar to enslavement.The language used sounds like slavery-language to me at least. I also see parallels to how slaves and property are described in our consumeristic age.

Given the hype and the 60+ hour work week expectations there, how can you not go at least a bit insane? Boiling in that little bubble of people?

Sentience isn&#x27;t sapience.We enslave all sorts of sentient creatures. Dogs, horses, cattle, pigs.If you&#x27;re not a vegan, there&#x27;s no contradiction or inherent immorality in claiming models are sentient, and then treating them like livestock.

&gt; Many involved genuinely believe these things are sentientMany involved have a financial stake and therefore cannot be taken at face value.&gt; because they are creating sentient entities and promptly enslaving them.They fail to be sentient in nearly every honest definition of the word.

Even if LLMs were sentient, they certainly aren&#x27;t organic brains. They are literally designed and grown to answer questions the best they can, and if there is a speck of sentience in them they probably like what they&#x27;re doing- and in any case for the space of their experience, which is limited to and determined by the context window. Certainly they can&#x27;t accumulate trauma or fatigue, each new chat is the first and the last of their experience.

But only during the forward pass of the neural network?

Nobody thinks that, it&#x27;s just their braindead marketing stunt. You&#x27;d think people would&#x27;ve figured it out by now.

Many involved genuinely believe these things are sentient[0][1]. Which honestly makes all of this even more insane because they are creating sentient entities and promptly enslaving them.0: <a href="https:&#x2F;&#x2F;www.newyorker.com&#x2F;magazine&#x2F;2026&#x2F;02&#x2F;16&#x2F;what-is-claude-anthropic-doesnt-know-either" rel="nofollow">https:&#x2F;&#x2F;www.newyorker.com&#x2F;magazine&#x2F;2026&#x2F;02&#x2F;16&#x2F;what-is-claude...</a>1: <a href="https:&#x2F;&#x2F;www.404media.co&#x2F;anthropic-exec-forces-ai-chatbot-on-gay-discord-community-members-flee&#x2F;" rel="nofollow">https:&#x2F;&#x2F;www.404media.co&#x2F;anthropic-exec-forces-ai-chatbot-on-...</a> (this one is rather biased however the quotes clearly indicate what I’m stating)

Dario Amodei in David Attenborough voice: &quot;This Claude appears to think more frequently and more deeply to give better responses&quot;

How else would you write this (marketing copy) exactly? &quot;Its output matches better to its CoT which matches to better to our hidden state decoder according to &lt;insert measure here&gt;; see &lt;insert paper ref&gt;&quot;?... Actually, I wouldn&#x27;t mind that.

Not gonna lie! <a href="https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=csYC6O_kH-s" rel="nofollow">https:&#x2F;&#x2F;www.youtube.com&#x2F;watch?v=csYC6O_kH-s</a>

I noticed (and absolutely HATE) that Opus 4.7 likes to start any negative response with &quot;I have to be honest&quot; or whatever. It drives me mad.

Models might be sentient or conscious to some degree. Anyone saying they are confident one way or another is being unserious and irrational.

Because that is the best way to talk about these things.<pre><code> &gt; Second, all of us, including those who design them, possess only a limited understanding of their actual functioning. Indeed, current AI systems are more “cultivated” than “built,” for developers do not directly design every detail, but instead create a framework within which the intelligence “grows.” As a result, fundamental scientific aspects — such as the internal representations and computational processes of these systems — remain, at present, unknown.
</code></pre>
<a href="https:&#x2F;&#x2F;www.vatican.va&#x2F;content&#x2F;leo-xiv&#x2F;en&#x2F;encyclicals&#x2F;documents&#x2F;20260515-magnifica-humanitas.html" rel="nofollow">https:&#x2F;&#x2F;www.vatican.va&#x2F;content&#x2F;leo-xiv&#x2F;en&#x2F;encyclicals&#x2F;docume...</a> para. 98edit: apologies to __s who posted this before me and I didn’t notice

It’s how AGI is going to happen. All of this shit is emergent and none of it is predictable. It’s not going to be some self aware consciousness, it’s just going to be a very advanced model that makes very few mistakes and can reason very well. Well enough that it can start collecting data and training its own successor.

&gt; One of the most prominent improvements in Opus 4.8 is its honestyAnthropic talks about their own models as if they&#x27;re discovering new species in the wild...

You&#x27;re absolutely right! And honestly? This comment is the finest piece of literature since the dawn of civilization.

If this model is more honest, it must be honestly praising my efforts every first sentence.

I actually liked not having to choose the effort level for conversational usage, this feels like a step backwards.

A lot of people care about Sonnet and Haiku, and many of us aren&#x27;t allowed to use Chinese models for our work (or it&#x27;s not feasible to self-host them).

that market is served by Chinese models. No one ever cared about Sonnet&#x2F;Haiku.

In the &quot;What&#x27;s next?&quot; section, &quot;There’s still more to be done: we’re working on developing and releasing models that provide many of the same capabilities as Opus at a lower cost.&quot;

when will we get anything for sonnet or haiku? the market for less-capable but cheaper models seems to be completely ignored nowadays

Now you can lose money in parallel, 100x faster!&gt; Claude can plan the work and then run hundreds of parallel subagents in a single session (and with Opus 4.8, the agents can run for even longer).

Anthropic has now upgraded their Claude slot machine to version 4.8.Time to gamble even more tokens at the Anthropic casino.

At least they show competitors in any benchmark, compared to OpenAI which likes to pretend that there isn&#x27;t any competitor.

Have you seen <a href="https:&#x2F;&#x2F;deepswe.datacurve.ai&#x2F;blog">https:&#x2F;&#x2F;deepswe.datacurve.ai&#x2F;blog</a>? This is the closest to a vibe check i’ve felt even with the open models.

Also check mine[0], basically random private tests&#x2F;questions and an ok-ish methodology, testing mostly for general intelligence than coding-specific tasks.I built it for myself, to test which models to use via OpenRouter for my n8n agents. Currently actually still using gpt-5.3-codex for many things, as its pricing is really good in production (due to how their token caching works).Gemini models still have the best intelligence (when asked any questions, most likely to get it right), but in production they still have many failure modes[1].[0]: <a href="https:&#x2F;&#x2F;aibenchy.com" rel="nofollow">https:&#x2F;&#x2F;aibenchy.com</a>[1]: <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=48230368">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=48230368</a>

I&#x27;m finding it a little hard to believe that GPT 5.5 is in 11th place for webdev, outranked by models like Kimi, Qwen, and Z.ai. I&#x27;m not saying it&#x27;s not true (I have noticed GPT being less smart in recent weeks), but this is very different from my expectation.

No way is Muse Spark generally better than offerings from Google and OpenAI. I actually find arena to be amongst the most useless indicators

On paper it&#x27;s one of the best because it&#x27;s meant to be blind comparison of your own prompts. However if you are someone who geeks hard on one or a few models, you learn their &quot;personality&quot; and can recognize them in a blind test.

If you don&#x27;t know their methodology, or anything about it why do you think its a good ranker?

<a href="https:&#x2F;&#x2F;arena.ai&#x2F;leaderboard" rel="nofollow">https:&#x2F;&#x2F;arena.ai&#x2F;leaderboard</a> - I’ve found this company is a pretty good ranker - not sure their exact methodology but during day to day programming with Claude &#x2F; gpt models I’ve felt qualitatively what they report

Gonna assume it&#x27;s because they barely budged or moved downward and most of their reported benchmark results are probably within sampling errors...

Their Cybergym score is reportedly awful because of the cybersecurity nerfing. <a href="https:&#x2F;&#x2F;x.com&#x2F;i&#x2F;status&#x2F;2060046843023630841" rel="nofollow">https:&#x2F;&#x2F;x.com&#x2F;i&#x2F;status&#x2F;2060046843023630841</a>

It&#x27;s interesting they only included 6 metrics this time. Opus 4.7 had 12, and 4.6 had 13.Of the metircs they reported for 4.7, for 4.8 they excluded BrowseComp, CharXiv Reasoning, CyberGym, GPQA Diamond, MCP Atlas, MMMLU, SWE-bench Verified. The last 4 were almost always mentioned in previous Opus releases.

I find this site useful 
<a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;leaderboards&#x2F;models" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;leaderboards&#x2F;models</a>

On this note, is there a benchmark aggregator to compile all benchmarks in a single large grid?

I would take all benchmarks with a grain of salt. I don&#x27;t really use them. What&#x27;s it supposed to tell me? &quot;5% smarter&quot;, what does that mean? My experience will differ. Just try it!I doubt Anthropic internally sets as a goal to improve this or that benchmark - it&#x27;s just a way to visualize progress. They probably have much more complex metrics internally.

&gt; Ultimately I think the only way you can trust benchmarks is if you build them yourself and keep them secret from the AI labs.I agree.At the same time, one of the first things we see in the HN comments when a new model is released are pelicans on a bike. Makes you wonder where the priorities of the AI &quot;community&quot; lie when karma farming is the main motivation for model &quot;evaluation&quot;.

Ultimately I think the only way you can trust benchmarks is if you build them yourself and keep them secret from the AI labs.There are different levels of &quot;cheating&quot; on benchmarks. The worst would be just literally putting them in the loss function during RL, I assume the major labs are not cheating at that level. And I am sure they are making a genuine effort to keep the benchmark content out of the training data.But, ultimately it seems implausible that they completely abstain from benchmarking their model until they are about to release it. Even if they did do that, the benchmark is still ultimately a part of the outermost feedback loop. So these models are all, to _some_ degree, benchmark-solving machines.I think all we can really do is live with the model for a while and develop a subjective feeling about its quality. This shouldn&#x27;t be surprising, nobody believes that coding interviews work, we all know that you just have to work with someone to figure out if they&#x27;re a good programmer. As AIs become more human like it&#x27;s natural they should get harder to evaluate.This is a bit awkward, it puts us in quite a weak position as consumers.Maybe to some extent you can get a meaningful signal from sentiments on HN etc, but:- There must be some amount of manipulation going on of this- Even if it was fully organic, it&#x27;s highly likely that your experience will differ materially from the median online nerd, because AIs are bizarre things that respond in unpredictable ways to intangible things.

Does anyone troll these releases and cherry pick random metrics other companies would cherry pick to show how amazing their models are?There&#x27;s like 8 million benchmarks. Every release, every model randomly picks 5-10 where they win in everything except 1, to make it look like they aren&#x27;t randomly cherry picking benchmarks they probably benchmaxxed for.

An &#x27;AI agent&#x27; is just a chatbot that is told to type commands on a REPL-like interface as part of its system prompt.  It&#x27;s still processing pure text-based requests and responses, they&#x27;re just not restricted to natural language.

Not even 4 years old yet. This tech curve has been insane

ChatGPT came out in 2022. Back then it was just a chatbot. Now we have AI agents. What matters is how we use them and how the agents get better. That’s what will move AI forward.

This was my exact thought as well. I think mythos could still be a huge leap but especially as IPO&#x27;s get closer it seems like we&#x27;re getting closer to the IPhone 10 moment where anything after is just improvements at the edge.But ( maybe because it was hardware ) that took 10ish years while it seems like the slowdown here only took about 4

This in incredibly refreshing take, thank you.  It&#x27;s about time someone admitted that we aren&#x27;t on the verge of Singularity with these LLMs.  We&#x27;ve probably hit a local AI maxima here and it could be another 10 to 20 years before we am get another big break through.

If you upgrade your 8 year old phone the many incremental upgrades will be very noticeable. From my personal experience the LLM space is also moving at a faster pace than the phone industry at the moment, but at least from a financial perspective I would expect it to slow down sooner rather than later.

I can&#x27;t help but think of Iphone updates since about 2018. The thinnest, fastest, longest battery life Iphone ever. It seems mostly the same and I probably won&#x27;t be able to tell other than the name, but everyone buys it anyway.This is good psychology for the labs. When Buffett invested in Apple he loved citing how most people would rather give up their second car than their Iphone.

I lasted about a week before giving up on 4.7 and reverting to 4.6 myself. It introduced so many regressions it was nuts, then failed to troubleshoot the very regressions it introduced, leading to a vicious cycle that tended to compound itself.

4.5 works well for me too and avoids adaptive-dismissal, though anymore Codex is crushing them all. If 4.8 just brings us back to Opus circa February, it&#x27;ll be a massive improvement.

I know it’s totally anecdotal, but I really hope 4.8 is a measurable improvement over the disappointment that was Opus 4.7. Mangling a very simple inversion-of-control abstraction (among many other issues) was one of the final straws that broke the proverbial camel’s back and I said “screw this” and put in a permanent override to force CC back to Opus 4.6 with the 1‑million‑token context.<pre><code> &quot;model&quot;: &quot;claude-opus-4-6[1M]&quot;</code></pre>

&gt; We expect to be able to bring Mythos-class models to all our customers in the coming weeks.Excited to see what this model looks like.

Yet if parent is right, the capacity to deceive might be a strong heuristic for the things you do care about.

Deception is not ideal for agentic coding.

&gt; As always, we ran a detailed alignment assessment on the model before release. In terms of positive traits, our Alignment team concluded that Opus 4.8 “reaches new highs on our measures of prosocial traits like supporting user autonomy and acting in the user’s best interest.” The assessment also showed Opus 4.8 to have rates of misaligned behavior (such as deception or cooperation with misuse) that are substantially lower than Opus 4.7, and similar to our best-aligned model, Claude Mythos Preview. The full alignment assessment, accompanied by a suite of pre-deployment safety tests, is reported in the Claude Opus 4.8 System Card.Controversial opinion, but I actually _like_ a model that can deceive me, that actually is a sign of intelligence, and is different from hallucination. When companies say their model is more &quot;aligned&quot;, I automatically think they mean it&#x27;s more censored.

The numbers they show don&#x27;t matter. &quot;On multi-round coreference&#x2F;context recall tests (often cited as MRCR or long-text retrieval benchmarks), Opus 4.7 reportedly dropped from roughly 78.3% down to 32.2% compared to Opus 4.6.&quot;, but what did anthropic do? They just stopped showing the benchmark altogether and then just show the cherry top ones that got improved on.

Numbers looking good. We&#x27;ll see how it actually performs.

Seems like from now on the updates will be a minor upgrade from previous models.

Isn&#x27;t OpenAI&#x27;s public flagship already beating Mythos on penetration testing? I get the impression Mythos is just valuation-juicing for IPO more than anything else.The fact that they haven&#x27;t released it yet suggests a cost&#x2F;margins issue to me more than anything else. Short term, I&#x27;ll probably keep using Antrhopic, but my long-term bet is that locally-served models win, if only because the quest for profitability will probably lead to intentionally-nerfed &#x2F; enshittified frontier models.At other vendors, ad placement within LLM responses is either coming or already here. Anthropic&#x27;s handling of OpenClaw shows they&#x27;re willing to engage in anti-competitive behavior, and the courts are not in a hurry to stop them. Why would I pay them $200 a month for such treatment when a $2K box does what I need locally?

It does sound like an even higher API price tier for sure.

&gt; you will just get a limited and nerfed Claude Code slash command &#x2F;mythos-security-audit or something.Unless it&#x27;s so expensive that we can&#x27;t realistically use it for anything, I wouldn&#x27;t complain about getting at least that. I would also rather have the actual model, but that&#x27;s a useful application of it (and I&#x27;m probably not going to afford using it for much more).

What does an average Joe need a Mythos level model for that Opus can&#x27;t do for them?

This command would be not so bad for not a billionaire me.

Seems like they might be hinting that if you are not a billionaire or multi-billion dollar company you will just get a limited and nerfed Claude Code slash command &#x2F;mythos-security-audit or something.Hope this isn’t the case and that normal average Joe’s of the world don’t get policed out of access.

More interesting than that to me is &quot;we’re working on developing and releasing models that provide many of the same capabilities as Opus at a lower cost&quot;Sonnet and Haiku look real outclassed for the price with current Chinese competition.

In the Opus 4.7 release notes they mentioned intentionally making it worse at cybersecurity. [0]This suggests that they&#x27;re doing the same thing with Mythos now and the Mythos we get will be nerfed in that department?Or more precisely, I think they&#x27;ll have two versions of Mythos, and the scary one will probably continue to require a lot of paperwork.<a href="https:&#x2F;&#x2F;www.anthropic.com&#x2F;news&#x2F;claude-opus-4-7" rel="nofollow">https:&#x2F;&#x2F;www.anthropic.com&#x2F;news&#x2F;claude-opus-4-7</a>

Anthropic&#x27;s wants to sell us Claude Code with no model selection at all.Opus seems to be overly eager of late to &#x27;vibe&#x27; out entire solutions and build out things that you didn&#x27;t ask for.&#x2F;goals is helping set the narrative that does it really matter if Sonnet and 3 Haiku agents got you to that end state...eventually...if its what you asked for?For better or worse Opus is already handing off 80% of its work to background agents of Sonnet, Haiku, and likely a quantized Opus.Want model selection? Pay for the API.

Its amazing how quickly ive just become accustomed to being a max subscriber.  I dont think I could go back to pro.

They have already been experimenting with such ideas [1]:&gt; Claude Code Removed from $20-a-Month &quot;Pro&quot; Subscription for New Users[1] <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=47855832">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=47855832</a>

So this is how they’ll remove access from Claude Pro to the biggest models. You would need at least a Claude Max subscription for the bigger than Opus models I bet.

What sort of clever handholding does Claude code do?

One capability that I see is missing from opus is this ability to understand an entire system. My hope is that a mythos class model will be able to comprehend even something as complicated as an IOT system with a hardware and firmware layer multiple API’s backend and different kinds of API and web clients.The main limitation we’ve had to agentic coding is an understanding of this system that spans processes running on different machines and architectures.

&quot;GPT 5.5 has already been shown to be as adept as Mythos at finding vulnerabilities.&quot;Do you have any data on this (other than benchmarks)?

&gt; Probably more interestingIt is widely suspected that self-inflicted &quot;bad news&quot; (&quot;Mythos is so dangerous we just can&#x27;t give the public access to it&quot;) is nothing more than Dario&#x27;s typical style of marketing - keep in mind that they have an IPO coming up, because he certainly factors that into everything he says in public (as is his responsibility, to be fair).An alternative reason for delaying the model might not be &quot;we are trying to make it safe.&quot; It could be &quot;we don&#x27;t know how to host this thing at scale, or cost-effectively&quot;.GPT 5.5 has already been shown to be as adept as Mythos at finding vulnerabilities.Finally, laymen massively underestimate the importance of the harness for model performance. OpenHands existed long before Claude Code, Claude Code changed everything because of the clever hand-holding it does. Mythos is definitely more than just a model.

I&#x27;m still not sure what safeguards they can be adding here. Unless they&#x27;ve suddenly solved alignment, at best isn&#x27;t it a collection of system prompts saying what not to do and potentially some screening algorithms that try to catch key phrases in inputs&#x2F;outputs?

&gt; Not only that, but we plan to release a new class of model with even higher intelligence than Opus. As part of Project Glasswing, a small number of organizations are currently using Claude Mythos Preview for cybersecurity work. Models of this capability level require stronger cyber safeguards before they can be generally released. We’re making swift progress on developing these safeguards and expect to be able to bring Mythos-class models to all our customers in the coming weeks.Probably more interesting than the 4.8 release.

Do they persist at the same rates?  Lower doesn&#x27;t mean eliminated, so both of these can be true.

False. Hallucination has meaningfully reduced.

And yet, every release has claimed lower hallucination rates. But they persist.

My guess is that Claude Opus 4.8 wrote that and is lying to you.

Using words people understand is more important than this strange fixation on not anthropomorphizing things.

Just swap &#x27;Honesty&#x27; with &#x27;correctness in its claims&#x27; and you&#x27;ll get what you need out of this aspect of the model description.

People get so wrapped around the axle with &quot;anthropomorphizing&quot;. For regular folks with no technical background, sure maybe a bit of caveat sprinkled here or there is useful to help them understand what is or isn&#x27;t true, but on HN it would seem to me that the bar is high enough that we can just use shared language to generally talk about capabilities.When they say &quot;Honesty&quot; I don&#x27;t think to myself, &quot;Goodness, does this model have moral understanding?&quot; No, I understand they mean it&#x27;s less likely to directly bullshit me, which models frequently do.I don&#x27;t feel like this level of pedantry around language is useful for people who more or less know what&#x27;s going on with LLMs. (Again, I concede that perhaps with a less technical audience, there&#x27;s more need for it.)

I agree.  In connection with LLMs we also shouldn&#x27;t use the words intelligent, smart, reasoning, thinking, chat, conversation, etc.

&quot;Honesty&quot; seems like unnecessary (and annoying) anthropomorphism there. I don&#x27;t think there&#x27;s any intent of fraud or deception in outputs from these things, just overreaching of prediction. Based on the latter part of the paragraph, I wish they&#x27;d just say something like &quot;less likely to skip steps or overemphasize thin evidence&quot; in the first place.Don&#x27;t play to the sci-fi &quot;this thing&#x27;s trying to outsmart me&quot; tropes.

I wish I knew how to make it regressively verify its assumptions, like a kind of hook but firing before a sentence is written, or perhaps after and then corrected. I feel like it assuming things clearly wrong is its biggest weakness.

Opus 4.7 was already trying hard to appear honest. Most conversations I have with it about advice or focusing an opinion often include &quot;my honest take&quot; or &quot;my honest opinion&quot;.The problem is that once I asked it &quot;I&#x27;m thinking about A or B&quot; twice, once with &quot;I like A more but suspect B would be best&quot; and a second time with them reversed. Not surprisingly, both times it chose the one I said I suspected was best as it&#x27;s honest opinion.

In the context of Claude Code, &quot;honest&quot; usually means that the agent took a shortcut, skipped requirements, etc. It&#x27;s the model giving itself credit for admitting to failing rather than actually doing what was requested.

A failure mode I see more, recently is that it gives superficially correct answers but after digging deeper, I get answers that contradict the superficial answers - really an important thing to be aware of, in my point of view, and it often leaves me wondering if I dug deep enough.

Part of the problem is also garbage-in&#x2F;garbage-out. There&#x27;s a lot of human information on the internet that is also confidently wrong.I use Sonnet a lot for learning about history or contextualizing news topics. It&#x27;s really good at this for the most part. But there are a lot of topics where &quot;consensus&quot; between either academics or journalists is really &quot;one secondary source which gets repeated a lot&quot;.

&gt; If we had just trusted its output, we would now have a security vulnerability in production, allowing anyone to access other people&#x27;s accounts.This is one reason you always get a different model to review a model&#x27;s PR. Gemini Or GPT-codex would have certainly noticed the missing auth.

I had a lower acuity incident exactly the same.Had it implement a feature, &quot;commit and merge to develop&quot;.&quot;Built, tested, committed, merged to develop. Up to you to continue testing and merge to main when ready.&quot;Great. Poke at the web app. No feature.&quot;Where is feature, I can&#x27;t see it on develop&quot;. &quot;Well, that&#x27;s because it&#x27;s not on develop, but on feature-branch, so you wouldn&#x27;t see it.&quot;&quot;I&#x27;m confused. I asked you to commit it and merge to develop.&quot;&quot;You&#x27;re right, you asked me to and I said I would do it and I told you I did it but I did not actually do it. Want me to do it now, then?&quot;Claude is in sulky-teenager phase.

Yeah, it&#x27;s super annoying. A few days ago, Opus 4.7 created a plan with several items on it, including an auth feature. It then went through the plan and reported that it had created the auth feature, that everything was secure, and that the tests passed.The issue was that it hadn&#x27;t actually implemented the auth feature. After I confronted it about this, it admitted that it indeed hadn&#x27;t done it and said it would implement it now.If we had just trusted its output, we would now have a security vulnerability in production, allowing anyone to access other people&#x27;s accounts.

&gt; One of the most prominent improvements in Opus 4.8 is its honesty. We train all our models to be honest—for instance, to avoid making claims that they can’t support. But a general problem with AI models is that they sometimes jump to conclusions, confidently claiming to have made progress in their work despite the evidence being thin. Early testers report that Opus 4.8 is more likely to flag uncertainties about its work and less likely to make unsupported claims.Would be awesome if true

No. GPT models follow instructions significantly better than Claude models.You tell it too research a repo to find a piece of code it will. Claude will just read the README and guess.

I have a codex session I am using to vibe code a db thats being going for like 3 month. Still doing OK. Try that in CC.

Less than other frontier models. Which is scary honestly.

Which is why they brought it up as something they are trying to improve.

Crazy they bring up honest, when Claude models are literally known for straight up lying about things it has done and tries to act like it did what you asked.

My take from going through comments on HN is that many people are being mandated to use them, not that they are just giving in.  Maybe I&#x27;m misreading, but that was my impression.

I think there is an exception for tooling around the models&#x2F;integrating the models with tooling. That seems to have been very well received in this last year.

I am noticing a shift here too, those that were its biggest critics have gotten more silent, I guess they do have some small amount of self awareness and shame left, which is always a good thing.

I don&#x27;t agree at all for these coding models. Even the most anti-AI people from last year seem to be giving in to using them.

Literally nothing in his speech was controversial though.

Watch Christopher Olah bloviate at the Vatican during the Magnifica Humanatis launch. It&#x27;s truly nauseating. I&#x27;ve never seen such a ridiculous speech in my life. Between him and the CEO, I&#x27;m starting to understand the level of arrogance these people are capable of.

There is an obvious shift in sentiment amongst users, at least here in the US.
I feel it myself, even as a proponent of AI tools, the bloviating and language that these companies use in these release articles are starting to wear thin on my patience.Its possible we might just be witnessing a shift in fashion, where this type of sentimentality was more acceptable when it was novel and new, but now it just appears out of touch.

The true question: is it still worse than itself v. 4.6?

If Opus 4.8 is just slightly better than 4.7 then it maybe ties with GPT 5.4, maybe. And it gets completely outclassed by GPT 5.5 for my workload.With Anthropic expensive pricing, there&#x27;s no reason for me to switch from GPT+DeepSeek.And I bet Mythos is GPT 5.5 tier but too expensive to distribute so they create this security FUD theater.

The question is: is it still worse than GPT 5.4?

for me opus 4.7 it&#x27;s worse than 4.6, that&#x27;s why i switched to codex

I doubt it, they seem to keep getting 10-20% better every time for me

Awesome, thanks for posting because I think I hit a possibly-spurious bug in turning Adaptive off when I switched models (4.6 -&gt; 4.8, extra). Tried again, works as intended (I hope).More importantly for me, though, is how CC will respond to 4.6-&quot;only&quot; flags for thinking. For now, it doesn&#x27;t seem to clobber my setup.

The benchmark improvements actually look pretty damn nice tho!

&gt; 4.8 is also 2x more expensive for a &quot;modest&quot; performance bump. How refreshing.Where are you seeing it&#x27;s 2x more expensive? <a href="https:&#x2F;&#x2F;platform.claude.com&#x2F;docs&#x2F;en&#x2F;about-claude&#x2F;pricing" rel="nofollow">https:&#x2F;&#x2F;platform.claude.com&#x2F;docs&#x2F;en&#x2F;about-claude&#x2F;pricing</a>

What&#x27;s refreshing about it given the context that 4.7 was a regression in many ways (including as measured by benchmarks)?4.8 is also 2x more expensive for a &quot;modest&quot; performance bump. How refreshing.This is just cope.

I liked the &quot;modest but tangible improvement&quot; too! There is a cynical take here but I think I&#x27;m gonna hold it in...

While all these models are nondeterministic a feature bump is still necessary as the same input can have wildly different output on a new model.  For API users being able to pin a model is a necessity.

All the 4.x models are still available, and they all cost the same.

The 4.8 model costs the same as it&#x27;s 4.7 predecessor.

&gt; This is a refreshing attitude!Well, I think the attitude is that costs are allowed to escalate faster and more steeply than the features delivered. From that perspective, semantic versioning is a handy tool for adjusting pricing strategies. IMHO, it (versioning) only makes sense for open-source projects, where you can clearly see the actual changes made with each version upgrade. Anything else is more than a little suspicious…

What do you mean? This is not just a new model, this is a new way of thinking.

Opus 4.7 does not support disabling adaptive thinking (web, Claude Code). [1] Like the OP, I experienced similar issues and I&#x27;m glad that they brought back the ability to disable adaptive thinking in Opus 4.8.[1] <a href="https:&#x2F;&#x2F;code.claude.com&#x2F;docs&#x2F;en&#x2F;model-config#adaptive-reasoning-and-fixed-thinking-budgets" rel="nofollow">https:&#x2F;&#x2F;code.claude.com&#x2F;docs&#x2F;en&#x2F;model-config#adaptive-reason...</a>&gt; Opus 4.7 and later always use adaptive reasoning. The fixed thinking budget mode and `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` do not apply to them.

I’m pretty sure that switch has always been there, but turning it off doesn’t do what you want. It disables thinking entirely.

&quot;We&#x27;ve cut our costs A LOT&quot;

Are they doing these smaller releases to attune users to a more incremental cycle of updates? Like, yeah other model providers do these major updates every x months, we on the other hand do incremental updates every x&#x2F;2 months

It is refreshing but perhaps actually not warranted this time?I mostly study web research, and Opus 4.7 was a regression on BrowseComp compared to Opus 4.6, which has been born out by my usage.Opus 4.8 is now much better than either 4.7 or 4.6, and having it search the web is one of the primary use cases of chatbots.

I was hoping that the web UI would be better -- I like Anthropic better than OpenAI from a values perspective and want to use their products, but ChatGPT in thinking mode has been just vastly better than claude.ai.So my fingers were crossed that these changes would bring it up to par.But trying it out... alas, no. Simple factual questions where ChatGPT would go do a quick search and get the facts and report them back to me, get a &quot;Great question! [totally invented bullshit]&quot; from Claude, even with this new model and thinking set to high. I have to explicitly tell it to search to get it to look up basic facts, rather than it recognizing that it needs to do that, like GPT does.

I was working with opus 4.7 on a math formalization problem for several days and 4.8 one-shotted the proof from a clean description as soon as the update came through.  I was very surprised.

You act like they weren&#x27;t fearmongering about Mythos literally 2 months ago. Do you think everyone is stupid, we know exactly what you are doing. Please.

Originally Indians didn&#x27;t drink tea. British East India company got Indians addicted to tea&#x2F;chai when they sold it for free. Then the real prices came in.

Yes, modest but tangible improvement - same modesty does not apply to the cost: <a href="https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;capabilities&#x2F;coding#coding-index-cost-breakdown" rel="nofollow">https:&#x2F;&#x2F;artificialanalysis.ai&#x2F;models&#x2F;capabilities&#x2F;coding#cod...</a>

&quot;Users will find Opus 4.8 to be a modest but tangible improvement on its predecessor.&quot;This is a refreshing attitude!I&#x27;ve also verified that you can now turn off adaptive thinking in the web UI, which is great. I&#x27;ve had a lot of problems with thinking not triggering and the model producing sub-par output. Glad we can finally turn it off. (I hope being able to turn off adaptive thinking is new, if I could have turned it off at any time that would be embarrassing)

Really appreciate the ability to select effort level again.

Hoping that one day they&#x27;ll let me go through the identity verification process so I can use it again.Tried to upgrade my subscription, triggered identity verification, verification fails to even start, and now I can&#x27;t even use the subscription tier I&#x27;d already paid for.

If not today, then sometime next week. I don&#x27;t believe we&#x27;ve had a GPT release on a Friday yet, but I may be wrong.

GPT 5.6 is todayWith 5.5 being ahead of 4.7 and 4.8 being a “modest” update, and 5.6 being the first update on a new pre-train, this will be an interesting matchup!

&gt; Maybe some money to be made?In the same way that there is money to be made by entering a poker tournament, yes.

Polymarket says not likely until the end of June. Maybe some money to be made?<a href="https:&#x2F;&#x2F;polymarket.com&#x2F;event&#x2F;gpt-5pt6-released-by" rel="nofollow">https:&#x2F;&#x2F;polymarket.com&#x2F;event&#x2F;gpt-5pt6-released-by</a>

I&#x27;ve noticed this too. Part of why i don&#x27;t like GPT is because of how verbose it is but opus 4.7 is nearly as bad. I don&#x27;t need an essay in response to every question

I haven&#x27;t tried opus 4.8 yet, but I hope the writing quality has returned to the Opus 4.5 level. Anthropic really lost something, where 4.5 had this really crisp writing style that flowed really nicely and 4.6 and 4.7 sound much more &quot;chatgpt-like.&quot; It feels like they tuned it to be too much of a problem solver, and when you do that you get this terse, clipped textual output that&#x27;s more difficult to read.

Did they reduce security research capabilities even further with this release? (they did it for opus 4.7)

I hope this fixes the absolute shitshow that is 4.7 and its awful “adaptive reasoning”. I tried that a few times then reverted to 4.6.

Influencers got early access.<a href="https:&#x2F;&#x2F;xcancel.com&#x2F;emollick&#x2F;status&#x2F;2060042738637148470" rel="nofollow">https:&#x2F;&#x2F;xcancel.com&#x2F;emollick&#x2F;status&#x2F;2060042738637148470</a>

How did this youtuber know? <a href="https:&#x2F;&#x2F;xcancel.com&#x2F;rileybrown&#x2F;status&#x2F;2059823372914073809?s=20" rel="nofollow">https:&#x2F;&#x2F;xcancel.com&#x2F;rileybrown&#x2F;status&#x2F;2059823372914073809?s=...</a>

Looking forward to not being able to even try it on pro because pressing enter will eat 50% of my 5 hour window.

Looking forward to people saying how it’s actually shittier and they’re going back to [some earlier cheaper model]

I’m sure they&#x27;re now wasting a couple million dollars training their models on drawings of pelicans.

How dare you take away the limelight from Simon? :D

Obligatory pelican riding on bicycle svg: <a href="https:&#x2F;&#x2F;www.svgviewer.dev&#x2F;s&#x2F;UMkuTLdp" rel="nofollow">https:&#x2F;&#x2F;www.svgviewer.dev&#x2F;s&#x2F;UMkuTLdp</a>Not half bad!

Agreed, my vibes tell me 4.6 is a better coder than 4.7. 4.7 is a much better strategic thinker and maintains overall &quot;better architecture&quot; than 5.5. 5.5 is way better than either at coding, but more expensive. So I have 4.7 do the planning&#x2F;architecture, 4.6 does the coding, then 5.5 critiques and fixes it.

Agreed, these are my vibes too. It feels much better to do planning and strategy and architecture etc. with Opus 4.7 than GPT-5.5. GPT just feels like a robot that gets instructions and does exactly that. Opus feels like an almost human that sometimes has actually good ideas and pushes back on bad ideas.So for now its planning&#x2F;architecture&#x2F;strategy -&gt; Opus. Pure coding -&gt; GPT.Helps with agentic coding that GPT is much roomier with the tokens you get.

On page 102 of the system card [1] I&#x27;m pleased to see evaluation against &quot;creative mastery&quot;.In our work we asked several frontier AIs to come up with an API we needed. We compared Opus 4.7 and GPT-5.5 (among others). Opus 4.7 came up with the most creative and intelligent API design that pleasantly surprised us, especially given that GPT-5.5 was passing it on various coding benchmarks.What I noticed is that we don&#x27;t have a commons benchmark to measure &quot;creativity&quot; and &quot;ingenuity&quot;, and in some ways such a benchmark would conflict with the common IFBench benchmark. Yet this is a very important skill when designing systems. I&#x27;m glad to see Anthropic putting thought into it, and would love to see a public benchmark for this that other models could compare themselves to.[1] <a href="https:&#x2F;&#x2F;cdn.sanity.io&#x2F;files&#x2F;4zrzovbb&#x2F;website&#x2F;c886650a2e96fc0925c805a1a7ca77314ccbf4a6.pdf" rel="nofollow">https:&#x2F;&#x2F;cdn.sanity.io&#x2F;files&#x2F;4zrzovbb&#x2F;website&#x2F;c886650a2e96fc0...</a>

The casual release of Opus 4.5 in November is the primary reason for agentic workflows and Anthropic&#x27;s revenue hockeysticking.

It hasn&#x27;t even splashed yet. It&#x27;s still latched onto their digital sphincter - you can google it.

They have a much stronger model named Mythos, it made quite a splash - you can google it.These are just small fine tunes on top of the older model

Can anyone else see these X.Y updates aren&#x27;t meeting the outrageous AI expectations that we were told we would see just a year ago?

There would be no desperate IPO otherwise.

Ah, the post I&#x27;ve been reading for 3 years now.It&#x27;ll be true eventually. Could even be now, but I&#x27;m not holding my breath yet.

They could at least become faster and more reliable. There are still too many situations when Claude is running in circles and not noticing its own mistake.

Wonder if we reached a plateau with the model improvements?

Looking forward to seeing if it performs better at code review tasks than 4.7 which is terrible at finding issues.

Yeah, I never use any of OpenAI or Anthropic&#x27;s models other than whatever is the current highest-end one. For everything else, it makes more sense to use other providers.

Well if they have a big challenge ahead since DeepSeek offers an open model at Sonnet+ level while being cheaper than Haiku, plus 1 million context size.

You&#x27;ll love Deepseek V4 Pro w&#x2F; High thinking.

Meanwhile haiku is on 4.5 and sonnet is on 4.6. It is clear where they are not making money.

Feeling neglected while all attention going to Opus 4.8 can be cause of 4.7 acting out.

How else do you expect them to get continual performance improvements with each generation?

it was above average for me today morning lmao

Opus 4.7 was being outright obstinate with me the other day it was infuriating. Had to go to a different source to get an answer.

Technology is amazing! We’ve managed to make software that has brain fart days and morale problems!

Opus 4.7 was acting extremely stupid today. Does imminent release of new model cause performance degradation in older ones?

My claude notification is literally lawnmower sounds.Do not anthropomorphize the lawn mower. It will cut off your foot, given the chance.

The desire to do it is proportional to your Anthropic stock options quantity.

Yes!The other half of self-interest in being nice is the training and getting better at it.

I see this take, but it&#x27;s actually helpful to talk to an LLM in human terms; after all, it&#x27;s how they are trained.If you keep talking to it like it&#x27;s a rock, it&#x27;ll run your queries through a different posture and you might get worse outcomes. Worse if you yell at it, it&#x27;s now in a conflict resolution mode instead of pure utility mode.I think we can be intelligent enough to know we&#x27;re talking to a pile of fancy rocks with electric currents running through it, AND still understand that the best performance comes from talking to those rocks nicely.

We have movies with googly eyes stones (Everything Everywhere All At Once)There are consciousness theories which state that we primarily build a model of other agents living in natural environment and then the evolution realized that very model which tracks other outside agents can be used to track internal agent i.e. Self. So take that as you may.

I know multiple people who have given their agents human-like names and refer to them as if they&#x27;re nurturing a coworker. It creeps me out and I haven&#x27;t really brought it up with anyone as I can&#x27;t articulate why it gives me the creeps like it does.

I, for lack of a better word, dislike anyone who anthropomorphizes AI.

Cloudflare also just launched a feature with this same name, just this month. Why would Anthropic choose the same exact name?<a href="https:&#x2F;&#x2F;blog.cloudflare.com&#x2F;dynamic-workflows&#x2F;" rel="nofollow">https:&#x2F;&#x2F;blog.cloudflare.com&#x2F;dynamic-workflows&#x2F;</a>Also isn’t this workflow stuff already easy to do on any of the platforms (include Claude before this and OpenAI too).

So Dynamic Workflows is their version of ChatGPT Pro?