Unfortunately, I don;t think that can ever be fixed. For an LLM to know that it is not hallucinating about something, it must know that that statement(s) is/are true. Which it cannot infer due to Godel's Incompleteness theorems.
You can either install that skill or put the Rules section directly in your Global CLAUDE.md for Claude or Personalization setting for Codex and it should cut down the output verbosity by quite a fair bit.
> 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.
It is widely suspected that self-inflicted "bad news" ("Mythos is so dangerous we just can't give the public access to it") is nothing more than Dario'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 "we are trying to make it safe." It could be "we don't know how to host this thing at scale, or cost-effectively".
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.
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.
It's interesting that (for example for the explore agent https://github.com/Piebald-AI/claude-code-system-prompts/blo... ) they use a personality "you are a file search specialist" and "your strengths" framing. I thought that was largely thought to be useless, or even counterproductive nowadays? Does anyone know more about this stuff?
More interesting than that to me is "we’re working on developing and releasing models that provide many of the same capabilities as Opus at a lower cost"
Sonnet and Haiku look real outclassed for the price with current Chinese competition.
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.
Anthropic's wants to sell us Claude Code with no model selection at all.
Opus seems to be overly eager of late to 'vibe' out entire solutions and build out things that you didn't ask for.
/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.
As long as they provide the same utility / $ I don’t see why not. It’s not like the open weight models are that far behind and Claude code itself shouldn’t be very hard for the commmunity to replicate if Anthropic start acting up too much.
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 /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.
> you will just get a limited and nerfed Claude Code slash command /mythos-security-audit or something.
Unless it's so expensive that we can't realistically use it for anything, I wouldn't complain about getting at least that. I would also rather have the actual model, but that's a useful application of it (and I'm probably not going to afford using it for much more).
Price discrimination is I think fine and reasonable so long if you can drum up the cash you can use it how you want within their ToS.
Although mental safety gymnastics aside, getting the most amount of intelligence for the cheapest amount of cost to normal people seems like the most ethical thing a big lab could do.
Going around and granting different tiers of intelligence to different insiders, friends, or companies is majorly problematic long-term.
Heck right now, the tokens you buy today for “Opus 4.8”, no one even knows or believes will be the same “Opus 4.8” just 3 days from now.
/security-review already exists so I don't think it would be crazy to have a /mythos-security-review as more thourough command as well. I think it's more likely it is going to be released at some point to the general public though - although the the pricing might make it quite unattractive.
It's not just better at cybersecurity, it's better at all the things (or most of them). I for one would really benefit from a better claude code. I still have to babysit it pretty closely to keep it from messing things up. Opus 4.7 was not an upgrade for me.
But in general, what does the average Joe need Opus for that Sonnet or Haiku can't do for them? Better is better.
Isn't OpenAI'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't released it yet suggests a cost/margins issue to me more than anything else. Short term, I'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 / enshittified frontier models.
At other vendors, ad placement within LLM responses is either coming or already here. Anthropic's handling of OpenClaw shows they'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?
Mythos is dramatically better specifically at finding zero-day vulnerabilities and developing exploits for them, that being what it was designed to do. On other cybersecurity tasks, GPT-5.5 is at least as good, but finding and exploiting zero-days is a particularly scary capability, which is why Mythos is a big deal. See, e.g., https://forum.effectivealtruism.org/posts/8yztpbjuPkyXsmA6n/....
We did not explicitly train Mythos Preview to have these capabilities. Rather, they emerged as a downstream consequence of general improvements in code, reasoning, and autonomy. The same improvements that make the model substantially more effective at patching vulnerabilities also make it substantially more effective at exploiting them.
I've been assuming that Mythos is just a big jump in model size, and that's where the jump in capabilities comes from. Hence I expect OpenAI not to be able to catch up without scaling up the model and hence significantly raising the API prices.
Anthropic frames this as something emergent. Not 100% but in a way they always phrase it as like, it’s a great model, but our breaths were swept and taken with its approach to security.
I'm still not sure what safeguards they can be adding here. Unless they've suddenly solved alignment, at best isn'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/outputs?
Because with Max subscriptions, you have to use the Claude Agent SDK, which is basically running Claude Code underneath. You don't get to use the chat/Messages APIs with personal subscriptions, for that you need the API pricing.
The point of a union is not prevent layoffs and role reduction in force. It is to make sure that the company has done its absolute best to find another solution to the problem (that they're trying to solve with layoffs) and ensures that the company is accountable for all their decisions.
This just limits hiring in the first place as we see in European branches of (generally, American) tech companies. And regardless, what solutions can be found by the union?
As for what solutions union provide - I have already explained. They exist to protect workers' interests. They do not exist to solve company's problems. That is the job of the company.
For hiring, I meant it in terms of a chilling effect in that companies will not hire as many people as they are much harder to fire.
You said "It is to make sure that the company has done its absolute best to find another solution to the problem," and I am asking, what solutions have unions "made sure" that the companies have done instead of layoffs? I am looking for concrete examples as I often hear this claim but I'm not sure what a union can do if the solution is layoffs, so my question is, what other solutions are there?
Then, I suggest that you educate yourself on how unions function and what they do. It's not my job to educate you. You can also ChatGPT this by the way.
I see, looks like you don't actually have any examples to give despite you yourself making such a claim, on a discussion forum and thread to discuss a topic, so lead with that next time, because that's all that the phrase "It's not my job to educate you" means.
Is that a serious question? Providing for education possibilities, legislating against noncompete agreements to hinder workers to start out on their own, challenging illegal decisions, the list goes on.
The claim was that the union would find solutions instead of a layoff so how would legislating help in that immediate case where a company wants to lay people off in the next quarter? I'm not asking about general things unions do, I'm challenging this specific claim the parent made.
A union's objective is to protect and safeguard the interest of its members. As long as it does that, it adds value, regardless of what anyone else (especially those not in the union) thinks whether it has value.
This thread is filled with so many anti-union takes that you have to wonder if they are paid bots.
If you assume that most of the crowd that visits this website works in tech or tech-adjacent fields, how can you be against an entity whose main objective is to safeguard and work for your interests over your employers? Unfathomable that people are willing to do so much against their own best interests. Or...they are bots.
Unions are complicated. Generally speaking they are good for workers, but when they start focusing on job security over compensation and build in seniority-based advantages, the leverage unions have can be wielded against consumers and fellow workers, not just against employers.
“Unions are unequivocally good” is about as naive as “unions are unequivocally bad.” It’s always a question of how the union prioritizes their power, and that can lead to bad practices in the long run.
If you care about fairness, generally, and not just “what’s good for me personally” you don’t have to look hard to see powerful unions acting in bad faith.
On the other hand, it's far more voluminous to catalog the list of corporations acting in bad faith and abusing their employees than finding the abuses of unions.
Firing managers for egregious behavior only makes the legal case for the victims. That's also why cities don't fire bad cops, but instead keep them around until pending litigation is resolved.
If “who is worse” is a relevant metric, the question of unions would not be complex. Again, though, this is an entirely naive view of what is a very complicated reality.
The very obvious reason that corporations are “worse” is simply that they have more leverage. The idea that “leverage is likely to be abused” is a much more thoughtful heuristic for the paradigm.
But unions have never existed in a vacuum. And without the context of why they came about, that is from corporations abusing employees, it's easy to say "Unions are complex" when the world in which they exist is far more complex than unions are and perhaps far more vile.
I'm generally pro-union. Don't think that just because I criticize them that I don't think they're generally a good idea. The problems I'm pointing to are general problems of democracy in general. Incumbents tend to ignore future generations well-beings when it comes to current generations ability to negotiate.
The point I'm trying to respond to is: "This thread is filled with so many anti-union takes that you have to wonder if they are paid bots."
I think there are plenty of reasons why normal folks are anti-union, and generally, it's because different sets of workers are in different positions and have different perspectives.
Generally speaking, if there were some kind of "Workers Bill of Rights" built into organized labor law, preventing these abuses, there would be much stronger support for unions generally.
You want to be a longshoreman? Tough shit, they aren't any jobs for you as a longshoreman... and it is a total coincidence that the extremely high paying gigs for longshoremen tend go to the children of existing longshoremen. Not to mention their effort to shut out technological improvements that are standard in most other countries now.
You want medical costs to go down? Tough shit, the professional organizations for medicine have managed to artificially limit the number of med school students and residencies.
If there were limits on what unions could to stifle competition within their own industries, if there were limits on the extent of job security for poorly performing union members, if there were legitimate rules that meritocracy has to be the rule, not the exception, then I think the vast majority of Americans would start clamoring for more union membership. What we currently see is a lot of good work, but also a lot of fiefdoms being established and locked down.
There are unions that protect workers from firm's abusive practices. There are also unions that protect lamplighters job from the "tyranny" of the electric light bulb, and make everyone poorer in the process.
No argument from me, other than: All those things can be also be said about corporations.
You brought up lamplighters and the electric lightbulb. It's worth noting that corporations had a cartel that prevented lightbulbs from running longer than a few years. And to some extent that still happens.
Big Clive did a good video on the Dubai Lamp a few years back, and why you can't buy them anywhere but in Dubai.
> The very obvious reason that corporations are “worse” is simply that they have more leverage. The idea that “leverage is likely to be abused” is a much more thoughtful heuristic for the paradigm.
If you enjoy thought terminating clichés, I suppose.
Corporations, in general, have a very different set incentives and ways they can wield power and ways that people outside of their power structures can interact with them.
It's the same issue when people try to claim a corporation having the power to do X is the same as a democratic government having the same power. It's not.
It’s a plausible take that the best government in human history (say Sweden) is worse than the worst corporation. Sweden had an active eugenics program.
Do you have a particular historic incident in mind? In what national context was this taking place?
There have been very few incidents where a union successfully defended its particularistic interest in a way that harmed the interests of working people generally in the long term. Employers' associations have characterized union activity this way even when the unions were objectively losing ground in every way (e.g. declining wage shares of output, declining membership, etc).
For structural reasons, unions are constantly faced with a choice between limiting their interests to the defense of some small section of workers alone (e.g. lamplighters, software developers, truckers) or expanding solidarity with ever wider sections of the working class. To generalize for the sake brevity, unions that go the former route tend to become very weak and get captured by employer interests. Only unions that go the latter route, which requires them to adopt a broader view of their struggle, have any chance of becoming strong.
For a clear historic example in how these diverging approaches can play out, see Eley's discussion of the knife grinders union versus the metal workers union in turn of the century Germany (around page 77).
The advent of parallel treatment of different workers in a shop the late 1970's is one of the main reasons we saw union membership start declining. Two-tiered bargaining structures started to show up as soon as there were difficult economic headwinds. When one segment of the union with some seniority is unwilling to make sacrifices for the younger generations, you end up with as system that is bad for non-trivial number of workers, and is literally the opposite of meritocratic. Solidarity only goes as far as "I've got mine" for many folks, and when that happens, the union as a way of protecting workers turns into a vehicle for extracting concessions at the expense of others.
Examples of ruinous two-tiered contracts include UFCW's 1978 contract, Teamsters' 1979 contract, UAW's 1979 contract and their 2007 contract.
No disagreement from me there. I am a stalwart financial and activist supporter of LaborNotes, which organizes across unions to eliminate two-tier contracts.
But two-tier contracts are the result of employer power. Employers always portray them as an inevitable outcome of the legal form of collective bargaining; the historical record cannot sustain this reading.
The solution to two tier is more powerful unions, which are inherently more democratic unions, because wage workers only have leverage if they act in concert.
> ...how can you be against an entity whose main objective is to safeguard and work for your interests over your employers?
This is like asking how people can be anti Google when Google's mission is to "organize the world's information and make it universally accessible and useful."
I personally lean pro-union, but it takes very little empathy to understand that the people who are anti-union don't believe that the unions will serve their stated purpose.
I think you underestimate the anti-union propaganda of the last century. Any working class individual against unions parrot the same surface-level talking points predicated on their own limited understanding of labor. In a way they are bots; the alive variety.
I don't think they're bots, since I've seen the sentiment on HN for a long time. Instead, it reminds me of the idea of people seeing themselves as "temporarily embarrassed millionaires". If you see yourself as being part of the capital class in the future and think strong unions hurt the capital class, one might be against them.
If not careful, it can come across as elitist: the notion that unions are for people who are replaceable—people that need unions.
I wonder though if those absolutely certain that they thrive in a meritocracy will have second thoughts if/when Corporate decides they're replaceable by AI.
You can do the math yourself if you so choose. If you don't want to or disagree with the results, well, it's not my job to educate you, as you've said.
Google makes a competing product to Claude's main product? So competing, in fact, that they have to ban Googlers from using Claude in order to get enough dogfooders.
reply