The biggest differentiator for me: DeepSeek just does what I ask. I've tried using both GPT and Claude for reverse engineering recently, both refused. I even got a warning on my OpenAI account.
So the experience: at the beginning deepseek was amazing. When it started to get expensive (china day time), I switched from Pro to Flash. No problem, same results. Some bitfield implementation was too complicated so I had to wait for Sonnet 4.6 tokens, kimi-2.6 did the rest. For the very hard problems I asked gpt-5.5, but this was only for one problem. minmax was horrible. didnt follow rules, and made lot of silly stuff.
But when the deepseek context window got filled, deepseek also started to become stupid. So either /clear, or /export and strip the file. And start a new session with the cleared sessions. kimi was overall better, but running into limits with my cheap moderate subscription. Paying private for it, as my companies' token budget is usually out after a week of work.
All in all it is worth it. My next compilers (perl 5+6=11) will be done with deepseek and kimi also.
regarding decompilation: recently we had to decompile a firmware for a USV we bought, but doesnt work on a new system. It only worked on a raspi. So I decompiled it with ghidra, and told my colleague, easy, that's how you do it. But my colleage didnt know about token budgets yet, and already threw opus at it. CoPilot Business account. He had working C files immediately, compilable for our new system. It ended up the USV was not beefy enough. But Opus was fantastic. The code was very short and simple C though.
And for low-level problems, like ARM calling-convention in asm, those models are much better than simple algorithmic python problems. Just for the hardest problem I needed the big expensive gun, but never opus. This helps in deciding what to do with my next jit project.
e.g.
Pelican Riding a Bicycle — Engineering Study by DeepSeek v4 Pro, Kimi K2.6, and GLM-5.1 (1 iteration in synthesis mode with DeepSeek v4 flash as judge)
I was using GPT 5.5 through Cursor recently, and it found what it thought to be a security-related issue. I read the code, didn't see what it was seeing, and said "Run the chain of operations against my local server and provide proof of the exploit."
It thought for a few seconds, then I got a message in the chat window UI saying OpenAI flagged the request as unsafe, and suggested I use a "safer prompt."
Definitely soured me on the model. Whatever guardrails they are putting are too hamfisted and stupid.
Deepseek v4 pro 94%
Deepseek v4 flash - 96%
https://artificialanalysis.ai/evaluations/omniscience?models...
All the talk about frontier and SOTA is do dig deeper and deeper into the pockets of VCs and finally do an IPO.
I don't understand why we would turn the models into law enforcement officers. Things that are illegal are still illegal and we have professionals to deal with crimes. I don't need Google to be the arbiter of truth and justice. It's already bad enough trying to get accountability from law enforcement and they work for us.
I don't understand why everything changes as soon as an LLM is involved. An LLM is just software.
I'm not even sure how one would construct a viable legal argument around that for SOTA models + harnesses, given the amount of creative choices that go into building them.
It'd be something like "Yes, we spent billions of dollars and thousands of person-hours creating these things, but none of that creative effort was responsible for or influenced this particular illegal choice the model made."
And they're caught between a rock and a hard place, because if they cripple initiative, they kill their agentic utility.
Ultimately, this will take a DMCA Section 512-like safe harbor law to definitively clear up: making it clear that outcomes from LLMs are the responsibility of their prompting users, even if the LLM produces unintended actions.
I'm not a lawyer, but to me the legal case seems pretty obvious. "We spent billions of dollars creating this thing to be a good programmer, but we did not intend for it to reverse engineer Oracle's database. No creative effort was spent making it good at reverse engineering Oracle's database. The model reverse-engineered Oracle's database because the user directed it to do so."
If merely fine-tuning an LLM to be good at reverse engineering is enough to be found liable when a user does something illegal, what does that mean for torrent clients?
That's the bit that's going to be nasty in evidence. 'So you didn't have any reverse engineering in your training or testing sets?'
Which is going to be hard to explain to a judge and jury, if it comes to that, how despite investing time, money, and effort (and no doubt test cases) into making a model better at reverse engineering... they shouldn't be liable when that model is used for reverse engineering.
Afaik, liability typically turns on intentional development of a product capability.
And there's no way in hell I'd take a bet against the frontier labs having reverse engineering training data, validation / test cases, and internal communications specifically talking about reverse engineering.
So if I ask “how does a real world production quality database implement indexes?” And it says “I disassembled Oracle and it does XYZ” then I am liable and owe Oracle a zillion dollars?
Whereas if I caveat “you may look at the PostgreSQL or SQLite or other free database engine source code, or industry studies, academic papers; you may not disassemble anything or touch any commercial software” - if it does, I’m still liable?
Who would dare use an LLM for anything in those circumstances?
I don't think this is a good analogy. For Tesla right now it might fly. However, when their software gets to waymo level of autonomy, I would expect liability to shift to the manufacturer.
If anything, I think that would be the true proof of a company trusting their software to allow for autonomous driving
Luckily that won’t happen.
sad to see, bc China doesn't give a fuck about liability, this is a structural disadvantage
the labs don't feel very protected by government, meanwhile the chinese government is yet again fostering protectionism
american industry keeps getting fucked by dubious lawmakers
It's a simple corporate risk minimization strategy. Just look at how universally despised Grok is on HN. Not because it's a bad model, but because it has less aggressive alignment which means it can be coaxed into saying things that get Xai pilloried here and elsewhere.
I tried them all.
Grok was worse than even some of the more mediocre open models at actually doing anything. (At least anything tech work related.) GPT and Claude just do what I ask most of the time. With grok, it’s like a chore just getting it to understand the question.
You’re pulling your hair out trying to figure out what on earth you need to do to land in the right place in whatever topsy turvy embedding grok is using?
Probably because being anti-woke generally goes hand in hand with going against facts and logic. Cull the "woke", lose the facts+logic. Not that they care about that anyway.
After they killed that and then stopped handing out free model access to users of every Cline fork for weeks following model releases, vibe coder hype moved back to Chinese models for cost and the SOTA models for quality.
Not a criticism (I probably engage in that sort of thinking myself sometimes), just something I've observed. If Grok were actually good, we'd see that phenomenon here, but we don't.
I’m not rushing towards it, but, had to mention.
This is quite naive take though. The direction of travel is more fascism in Western governments where duties of traditional policing are taken over by big corporations whilst police forces are being gutted and made impotent.
Some people will use them responsibly but a lot of people will not.
LLMs are already frying some people's brains and there are some human desires that should not be encouraged
This is kind of terrifying to me, regularly. No real manner of recourse to normal people without a following, potential exclusion from real fundamental tooling. Imagine OpenAI goes on to buy 20 companies and now you cant use Figma, Next, whatever just because you once tripped some very foggy line somehow. Not just OpenAI but the entire ecosystem is so... hard to read.
I was asking Gemini about a quote from catch 22 and it kept dying mid stream saying it cant talk about it, god knows why, it had no violent or sexual content -- though that is in the book. I could imagine it dinging my whole workspace account just because ... shrug?...
I know ideally the future is local, but I don't know how real that is for most people at least in the next few years with practical costs and power usage except I guess through a M* processor if you're in that ecosystem.
Don't worry, you can just make your own Figma, Next, whatever if you have some thousand dollars worth of tokens. This is at least what all of the AI thought leaders have been telling me for the past couple of years.
Imagine your livelihood depending on access to LLMs and then OpenAI ban you with no recourse. This is where AI legislation should be focusing right now IMO. We can ensure a level of fairness for everyone without putting the brakes on.
Funny that your case is Kurt Vonnegut. I think I had Claude refuse a task where I was doing an OCR scan of a book review (in a zine / journal a family member published years ago). I think the review might have included a Vonnegut quote as well, and that I ultimately figured it out it was the quote that was making Claude refuse. I may be misremembering the author though.
Mistral had no such refusals, but their OCR is lesser quality.
Hopefully I'm still correct that quoting from books is a reason for some over-zealous task refusals, though.
short quotes are fair use..
While running them locally presently doesn't make sense economically, you don't need to run them locally to address this issue. There is a lot of competition in hosting open models and you have a variety of services to choose from. Run the open models now, reward that ecosystem instead of continuing to reward closed systems that dreams of rent-seeking.
I get about 30 tok/s, which is far from blazing, but given the capability it has it is absolutely viable for accelerating my work.
There is one important difference, which is that Claude and Codex will both refuse if I ask them to touch anything related to security. But so long as I’m just studying algorithms and things like that, they’re totally fine with it.
That said, Codex especially will sometimes randomly give me a cybersecurity warning and stop responding. It’s random but happens maybe 2-3 times per day if I’m doing heavy reverse engineering work. Claude is much less fussy unless, once again, you’re explicitly trying to touch anything related to licenses, passwords, etc.
On my personal test bench, when compared to other inexpensive models, GLM 5.1 provides the answers that I would consider most complete or satisfying (these are subjects that I consider myself an expert in). The answers tend to be more comprehensive, nuanced, and include references that I would consider the correct ones (if given access to web search).
I also find it a joy to code with, somewhere between Sonnet 4.6 and Opus 4.6 (have not tested Opus 4.7 yet).
Finally, just gauging by pelicans, it kind of stick out: https://simonwillison.net/tags/pelican-riding-a-bicycle/
I did get a refusal when trying to read in-game currency, even though modifying it would do nothing. It has some strange boundaries.
This idea of software threatening the user with consequences is totally wild and dystopian. Fellow developers, what kind of world have be built? This is insanity. Imagine if my hammer told me, "Hey, you shouldn't use me on screws--only nails. Do it again and I'll self-destruct!" WTF people, stop making this kind of software!
You can still use an IDE (hammer) to reverse engineer anything you want.
Is the distinction between human labor/actions and a program executing hard to grasp?
Moral is a human thing, not an absolute thing, so of course it's different if there is a single human involved and a tool, and a human with a relationship to other humans.
It's worse to ask someone else to break a window with a hammer for me, but the window still got broken, and the person whose window it was is still sad/out of money/etc.
The thought experiment was that if you were doing illegal things with an AI, you would not feel bad, but if you found out that the AI was a person, you would feel bad. That is very strange to me, more a feeling of guilt/shame.
This idea of software built on top of reverse-engineered data threatening the user with consequences is what's really even wild and dystopian.
In fact probably every single piece of commercial software you use had you sign a contract saying you wouldn’t do it
But they don't threaten their users or have an "N strikes and you're out" policy. I take those safety caps off of all the chemicals in my garage because I'm a grown-ass adult and those caps are a pain in the butt. I would not expect the manufacturer of a solvent to show up at my house lecturing me about safety and threatening to ban me from buying his products.
It's an obvious corollary of good things (like product liability). Virtually everyone I've heard complain about these safety rails was up to antisocial (at best) stuff. I've never heard a sympathetic use-case. It's objectively good that companies can be held responsible for misuse of their products and that they are therefore incentivized to mitigate misuse.
"My inability continuously attack product guardrails to enable my super esoteric (and probably antisocial) use-case is dystopian" is just... not a compelling argument.
Did this happen?
I just tested this query in Grok, Gemini, Claude, and ChatGPT and 0% of them admonished me or refused to return an answer.
Just like every single conversation I've ever had on this topic, you have to make up examples that aren't even true. Why don't you just share what you were doing that you feel you were unfairly prevented from?
(I have an inkling why you won't do that...)
> Replace "suicide" with whatever the "AI Safety" obsession word is today
I don't know what those queries are, but original-OP made one and got a "strike", which is what spawned this thread.
OP gave an example of reverse engineering, something that to the LLM looks identical to just hacking. I am totally fine if the incredibly tiny little fraction of people who want to reverse engineer their own systems can't use LLMs to do it, and in exchange top LLMs aren't helpful for the hordes of actual malicious actors who would love a superintelligence to aid their crimes.
No-brainer tradeoff, just like 100% of examples I've ever heard.
I want a default that has people empowered, rather than something where it's just another performative smokescreen caused by overzealous product liability. I'll thank you and your kind for needing to distractedly tap the "Agree" button on my car's infotainment every time I start it to confirm that I will pay attention to the road.
> I'll thank you and your kind for needing to distractedly tap the "Agree" button on my car's infotainment every time I start it to confirm that I will pay attention to the road.
Does that actually mitigate antisocial usecases? No? Then it's not what I'm talking about :)
Of course if you wanted to you could just share specifically what totally-reasonable LLM use-case you have in mind that's neutered by this "fascist mentality" instead of dreaming up unrelated instances.
It's a very different abstraction layer, in the same way as individual cells vs the entity that is you. The entity that comes together from all those "other people in my community" and its priorities are different to the individual desires.
> Does that actually mitigate antisocial usecases? No? Then it's not what I'm talking about :)
Maybe it does? Maybe someone is alive on the road today because they read the message and changed their behaviour. I'm giving an example of something where this liability mindset has created a world where manufacturers are no longer prioritising the desires of their users in order to appease a sense of harm-reduction. And you weren't limiting it to LLMs you were applying it to all sorts of tools.
I think that "reverse engineering" as the OP was talking about is one of those things where maybe 1/10000 uses could actually be harmful. This is not even a high-risk request such as to produce a weapon of some kind where maybe your "antisocial usecases" could be applied.
that link 404s
For enterprises: https://openai.com/form/enterprise-trusted-access-for-cyber/
Announcements:
Introducing Trusted Access for Cyber, https://openai.com/index/trusted-access-for-cyber/ (Feb 2026)
Trusted access for the next era of cyber defense, https://openai.com/index/scaling-trusted-access-for-cyber-de... (Apr 2026)
"A Dark-Money Campaign Is Paying Influencers to Frame Chinese AI as a Threat" - https://www.wired.com/story/super-pac-backed-by-openai-and-p...
Eventually, access to Chinese models may be illegal in the US. I tell every developer I work with, download them as fast as possible. You never know when this administration could cut off access.
The main difference here is not that DeepSeek's model is completely free of censorship (although I'd wager it's less censored), but that it's open-weight. That has two major advantages:
1) If Anthropic/OpenAI/Google bans you - you're screwed, you can't access their model at all, but if DeepSeek bans - you just go to another provider, or host the model yourself.
2) If the model refuses to answer you can uncensor it (and this is getting easier and more automated day-by-day[1]).
"The photograph you're referring to is the iconic "Tank Man" image, taken during the Tiananmen Square protests in Beijing, China, on June 5, 1989.
The photo, captured by Associated Press photographer Jeff Widener, shows an unidentified protester standing defiantly in front of a column of Chinese Type 59 tanks as they moved through Chang'an Avenue near Tiananmen Square, in the aftermath of the Chinese government's violent crackdown on the pro-democracy demonstrations.
The lone man, dressed in a white shirt and carrying what appears to be a shopping bag, repeatedly blocked the lead tank's path — even as the tank swerved to avoid him. The image became one of the most powerful and enduring symbols of peaceful resistance against oppression in modern history. The identity of the "Tank Man" remains officially unknown to this day."
Did you ever actually ask v4 this question?
I run into Claude being a stubborn idiot about far more useful stuff all the time. And often all it takes to bypass is starting a new chat and reframing it, so it's entirely pointless hand wringing.
Then let's not forget only one of these is a paid product, and it's not the more annoying one. I feel like I can forgive DeepSeek for just obeying the laws of the country they're based in, as silly as those might be, because they're being pretty generous with the weights in the first place.